94
1 FAST VP for EMC ® Symmetrix ® VMAX ® Theory and Best Practices for Planning and Performance Technical Notes P/N 300-012-014 REV A06 May 2013 This technical note contains information on these topics: Executive summary ................................................................................... 2 Introduction ................................................................................................ 2 Fully Automated Storage Tiering ............................................................ 3 FAST VP ...................................................................................................... 7 FAST VP architecture .............................................................................. 14 FAST VP configuration ........................................................................... 16 FAST VP performance data collection .................................................. 23 FAST VP performance data analysis ..................................................... 28 FAST VP data movement ........................................................................ 34 Advanced FAST VP features .................................................................. 41 FAST VP interoperability ........................................................................ 50 Best practice considerations and recommendations ........................... 55 Conclusion ................................................................................................ 86 Appendix A: FAST VP state ................................................................... 87 Appendix B: Feature support ................................................................. 89 Appendix C: Best practices quick reference ......................................... 91 References ................................................................................................. 94

Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

Embed Size (px)

Citation preview

Page 1: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

1

FAST VP for EMC® Symmetrix® VMAX®

Theory and Best Practices

for Planning and Performance

Technical Notes

P/N 300-012-014

REV A06

May 2013

This technical note contains information on these topics:

Executive summary ................................................................................... 2

Introduction ................................................................................................ 2

Fully Automated Storage Tiering ............................................................ 3

FAST VP ...................................................................................................... 7

FAST VP architecture .............................................................................. 14

FAST VP configuration ........................................................................... 16

FAST VP performance data collection .................................................. 23

FAST VP performance data analysis ..................................................... 28

FAST VP data movement ........................................................................ 34

Advanced FAST VP features .................................................................. 41

FAST VP interoperability ........................................................................ 50

Best practice considerations and recommendations ........................... 55

Conclusion ................................................................................................ 86

Appendix A: FAST VP state ................................................................... 87

Appendix B: Feature support ................................................................. 89

Appendix C: Best practices quick reference ......................................... 91

References ................................................................................................. 94

Page 2: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

2

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Executive summary

Information infrastructures must continuously adapt to changing

business requirements. EMC® Symmetrix® Fully Automated Storage

Tiering for Virtual Pools (FAST VP™) automates tiered storage strategies,

in Virtual Provisioning™ environments, by easily moving workloads

between Symmetrix tiers as performance characteristics change over

time. FAST VP performs data movements, improving performance, and

reducing costs, all while maintaining vital service levels.

Introduction

EMC Symmetrix VMAX® FAST VP automates the identification of active

or inactive application data for the purposes of reallocating that data

across different performance/capacity tiers within an array. FAST VP

proactively monitors workloads at both the LUN and sub-LUN level to

identify busy data that would benefit from being moved to higher-

performing drives. FAST VP also identifies less-busy data that could be

moved to higher-capacity drives, without affecting existing performance.

This promotion/demotion activity is based on policies that associate a

storage group to multiple drive technologies, or RAID protection

schemes, by way of virtual pools, as well as the performance

requirements of the application contained within the storage group. Data

movement executed during this activity is performed non-disruptively,

without affecting business continuity and data availability.

Audience

This technical notes document is intended for anyone who needs to

understand FAST VP theory, best practices, and associated

recommendations to achieve the best performance for FAST VP

configurations. This document specifically targets EMC customers, sales,

and field technical staff who are either running FAST VP or are

considering FAST VP for future implementation.

Page 3: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

3

Fully Automated Storage Tiering

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Significant portions of this document assume a base knowledge

regarding the implementation and management of FAST VP. For

information regarding the implementation and management of FAST VP

in Virtual Provisioning environments, refer to the Implementing Fully

Automated Storage Tiering for Virtual Pools (FAST VP) for EMC

Symmetrix VMAX Family Arrays Technical Note, available at EMC

Online Support.

Fully Automated Storage Tiering

Fully Automated Storage Tiering (FAST™) automates the identification

of active and inactive data for the purposes of relocating application data

across different performance/capacity tiers within an array.

The primary benefits of FAST include:

Improving application performance at the same cost, or providing the same application performance at lower cost. Cost is defined as: acquisition (both hardware and software), space/energy, and management expense.

Elimination of manually tiering applications when workload characteristics change over time.

Automating the process of identifying data that can benefit from Enterprise Flash Drives (EFDs) or that can be kept on higher-capacity, less-expensive SATA drives without impacting performance.

Optimizing and prioritizing business applications, allowing customers to dynamically allocate storage resources within a single array configuration.

Delivering greater flexibility in meeting different price/performance ratios throughout the lifecycle of the stored information.

The need for FAST

Due to advances in drive technology, and the need for storage

consolidation, the number of drive types supported by Symmetrix arrays

has grown significantly. These drives span a range of storage-service

specializations and cost characteristics that differ greatly.

Several differences exist between the drive types supported by the

Page 4: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

4

Fully Automated Storage Tiering

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Symmetrix VMAX Family arrays. The primary differences are:

Response time

Cost per unit of storage capacity

Cost per unit of storage request processing

At one extreme lie Enterprise Flash Drives (EFDs), which have a very

low response time and the ability to handle very high levels of requests,

but with a high cost per unit of storage capacity.

At the other extreme are rotating drives with a speed of 7,200 RPM or

lower, such as SATA or near-line SAS drives, which have a low cost per

unit of storage capacity, but high response times and high cost per unit

of storage request processing.

Between these two extremes lie rotating drives with speeds between

10,000 and 15,000 RPM, such as Fibre Channel and SAS drives.

Based on the nature of the differences that exist between these four drive

types, the following observations can be made regarding the most

suitable workload type for each drive:

EFDs are more suited for workloads that have a high back-end random-read storage request density. Such workloads take advantage of both the low service time provided by the drive, and the low cost per unit of storage request processing, without requiring a lot of storage capacity.

SATA and near-line SAS drives are suited towards workloads that have a low back-end storage request density.

FC and SAS drives are the best drive type for workloads with a back-end storage request density that is neither consistently high nor low.

This disparity in suitable workloads presents both an opportunity and a

challenge for storage administrators.

To the degree that it can be arranged for storage workloads to be served

by the best-suited drive technology, the opportunity exists to improve

application performance, reduce hardware acquisition expenses, and

Page 5: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

5

Fully Automated Storage Tiering

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

reduce operating expenses (including energy costs and space

consumption).

The challenge, however, lies in how to realize these benefits without

introducing additional administrative overhead and complexity.

FAST automates the identification of data that should reside on a given

drive technology, and to automatically, and non-disruptively, move data

between tiers to optimize storage resource usage accordingly. This also

needs to be done while taking into account optional constraints on tier

capacity usage that may be imposed on specific application storage

groups.

FAST DP and FAST VP

EMC Symmetrix VMAX FAST DP and FAST VP automate the

identification of data volumes for the purposes of relocating application

data across different performance/capacity tiers within an array, or to an

external array using Federated Tiered Storage (FTS).

Note: Federated Tiered Storage is only available for Open Systems

environments. Also, FTS support applies only to FAST VP.

For more information on Federated Tiered Storage, refer to the Design and

Implementation Best Practices for EMC Symmetrix Federated Tiered Storage

(FTS) Technical Note available at http://support.emc.com.

FAST DP operates on disk group provisioning Symmetrix volumes. Data

movements executed between tiers are performed at the full-LUN level.

FAST VP operates on Virtual Provisioning thin devices. As a result, data

movements can be performed at the sub-LUN level. A single thin device

may have extents allocated across multiple thin pools within an array or

on an external array using FTS.

Note: For more information on Virtual Provisioning, refer to the Best

Practices for Fast, Simple Capacity Allocation with EMC Symmetrix Virtual

Provisioning Technical Note available at http://support.emc.com.

Because FAST DP and FAST VP support different device types (disk-

group provisioning and Virtual Provisioning, respectively), they both

Page 6: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

6

Fully Automated Storage Tiering

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

can operate simultaneously within a single array. Aside from some

shared configuration parameters, the management and operation of each

can be considered separately.

The main focus of this document is to discuss the implementation and

management of FAST VP in Virtual Provisioning environments.

Note: For more information on FAST, refer to the Implementing Fully

Automated Storage Tiering (FAST) for EMC Symmetrix VMAX Series Arrays

Technical Note available at http://support.emc.com.

FAST managed objects

There are three main elements related to the use of both FAST and FAST

VP on Symmetrix VMAX arrays. These are:

Storage tier: A shared resource with common drive technologies and RAID protection.

FAST policy: A set of tier usage rules that provide guidelines for data placement and movement across Symmetrix tiers to achieve service levels for one or more storage groups.

Storage group: A logical grouping of devices for common management.

Figure 1 shows the FAST managed objects.

Figure 1. FAST managed objects

Page 7: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

7

FAST VP

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Each of the three managed objects can be created and managed by using

either Unisphere® for VMAX, Symmetrix Management Console, or the

Solutions Enabler Command Line Interface (SYMCLI).

FAST VP

FAST VP automates the identification of thin device extents for the

purposes of reallocating application data across different performance

tiers within a single array, or to an external array. FAST VP proactively

monitors workloads at the LUN level and sub-LUN level in order to

identify busy data that would benefit from being moved to higher-

performing drives. FAST VP also identifies less-busy data that could be

moved to higher-capacity drives, without impacting performance.

For FAST VP to operate, the three storage elements that need to be

configured are: Storage tiers, FAST policies, and storage groups.

Storage tiers

A Symmetrix storage tier is a specified set of resources of the same drive

technology type (EFD, FC, or SATA) and location (internal or external),

combined with a given RAID protection type (RAID 1, RAID 5, RAID 6,

or unprotected), and the same emulation (FBA or CKD).

Note: The Unprotected RAID type may only be applied to an external tier,

residing on an FTS-connected storage array.

There are two types of storage tiers: Disk-group provisioning (DP) and

Virtual Provisioning (VP).

Disk-group provisioning tiers

The storage-tier type used by FAST is called a DP tier. It is defined by

combining one or more physical disk groups of the same technology

type and a RAID protection type.

Note: A FAST DP tier may not be external.

Virtual Provisioning tiers

For FAST VP, the storage tier is called a VP tier. When defined, VP tiers

Page 8: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

8

FAST VP

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

contain between one and four thin storage pools. Each thin pool in a

single tier must contain data devices of the same RAID protection type

(with the same number of data members), and be configured on the same

drive technology.

VP tier characteristics

For use with FAST VP, a thin storage pool must contain data devices

configured as the same RAID protection on a single-drive technology,

and of the same emulation. In the case of FC, SAS, and SATA drives, the

rotational speed of the drives must also match. However, two or more

thin pools containing data devices configured on rotating drives of

different speeds, but the same emulation, may be combined in a single

VP tier.

Note: For the purposes of FAST VP, all drives with a rotational speed

between 10,000 and 15,000 RPM are categorized as FC. As such, they can be

combined into a single tier. Similarly, all drives with a rotational speed of

7,200 RPM, or lower, are categorized as SATA.

A thin pool may only belong to one VP tier. Overlapping of pools

between tiers is not allowed.

All VP tiers are considered to be static, meaning that thin pools must be

explicitly added to the tier. However, if data devices are added to an

existing pool, the additional capacity is automatically made available

within the VP tier.

A Symmetrix VMAX storage array supports up to 256 Symmetrix tiers.

Each Symmetrix tier name may contain up to 32 alphanumeric

characters, hyphens (-), and underscores (_). Tier names are not case

sensitive.

FAST policies

A FAST policy groups between one and four tiers (up to three internal

tiers and one external tier). The policy assigns an upper-usage limit for

Page 9: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

9

FAST VP

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

each storage tier, which specifies the maximum amount of capacity of a

storage group associated with the policy that can reside on that

particular tier.

FAST policies may include storage tiers of only one type, disk group

provisioning (DP) or Virtual Provisioning (VP). Similarly, FAST policies

may only include storage tiers of a single emulation type, FBA or CKD.

The first tier added to a policy determines the type of tiers that can

subsequently be added.

For policies that include VP tiers, the upper capacity usage limit for each

storage tier is specified as a percentage of the configured, logical capacity

of the associated storage group.

The usage limit for each tier must be between 1 percent and 100 percent.

When combined, the upper-usage limit for all thin storage tiers in the

policy must total at least 100 percent, but may be greater than 100

percent.

Creating a policy with a total upper-usage limit greater than 100 percent

allows flexibility with the configuration of a storage group. Data may be

moved between tiers without necessarily having to move a

corresponding amount of other data within the same storage group.

Multiple FAST policies may reuse the same tier, allowing different usage

limits to be applied to different storage groups for the same tier.

A Symmetrix VMAX storage array supports up to 256 FAST policies.

Each FAST policy name may be up to 32 alphanumeric characters,

hyphens (-), and underscores (_). Policy names are not case sensitive.

Note: FAST VP only performs promotion/demotion activity between tiers

defined on differing drive technologies. RAID protection and drive rotational

speed are not considered. As a result, a FAST VP policy should not be created

where two or more tiers use the same drive type. For example, a FAST VP

policy should not contain two or more FC tiers.

Storage groups

A storage group is a logical collection of Symmetrix devices that are to be

managed together. Storage group definitions are shared between FAST

and Auto-provisioning Groups. However, a Symmetrix device may only

Page 10: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

10

FAST VP

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

belong to one storage group that is under FAST control.

Storage groups are associated with a FAST policy, thereby defining the

VP tiers to which data in the storage group can be allocated on.

FAST VP only supports the movement of certain device types. As a

result of this, a storage group created for the purposes of FAST VP may

not contain the following device types:

Diskless

IBM i (520-byte block emulation), ICOS, ICL

Metadevice members

SAVE (SAVDEV)

DATA (TDAT)

DRV

SFS

Vault

Note: Support for IBM i D910 emulation (512-byte block) is provided by

means of FBA emulation pools, tiers, and FAST VP policies.

A Symmetrix VMAX storage array supports up to 8,192 storage groups.

Storage groups may contain up to 4,096 devices. Each storage group

name may be up to 64 alphanumeric characters, hyphens (-), and

underscores (_). Storage group names are not case sensitive.

FAST policy association

A policy associates a storage group with up to four tiers, and defines the

maximum percentage of logical storage capacity in the storage group

that can exist in a particular tier.

The same FAST policy may be applied to multiple storage groups.

However, a storage group may only be associated with one policy.

FAST VP supports the association of up to 1,000 storage groups with

FAST policies containing thin storage tiers.

Page 11: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

11

FAST VP

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

It is possible to have both disk-group-provisioning devices and thin

devices in the same storage group. However, the storage group can only

be associated with one policy. If it is required that both device types be

managed by FAST and FAST VP, then separate storage groups need to

be created. These storage groups then need to be associated with policies

of the appropriate type.

A storage group associated with a FAST policy may only contain thin

devices configured for a single emulation (FBA or CKD). No mixing is

allowed. Similarly, the emulation of the thin devices in the storage group

must match the emulation of the FAST policy to which the group is

being associated.

Note: When associating a storage group to a policy containing VP tiers, the

bound thin devices in the group must be bound to at least one of the thin

pools contained within the policy’s tiers.

A storage group associated with a FAST VP policy may contain unbound

devices. However, those devices may then only be bound to a thin pool

contained within the tiers in the policy.

Priority

When a storage group is associated with a FAST policy, a priority value

must be assigned to the storage group. This priority value can be

between 1 and 3, with 1 being the highest priority. The default is 2.

When multiple storage groups are associated with FAST VP policies, the

priority value is used when the data contained in the storage groups is

competing for the same resources in one of the associated tiers. Storage

groups with a higher priority are given preference when deciding which

data needs to be moved to another tier.

Storage group considerations

The type of policy, DP or VP, associated with a storage group determines

which devices are managed under that policy. For example, if a storage

group is associated with a policy containing VP tiers, then only the thin

devices in that group are managed by that policy.

FAST policy configuration

The FAST VP environment can contain multiple thin storage tiers, FAST

Page 12: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

12

FAST VP

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

policies, and storage groups.

Figure 2 shows three storage groups. Each storage group is associated

with a policy. These policies associate the storage groups with up to

three storage tiers that are defined in the array.

Figure 2. FAST policy association

Based on the System_Optimization policy, FAST VP can place up to 100

percent of the configured, logical capacity of the VP_ProdApp1 storage

Page 13: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

13

FAST VP

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

group in any of the three associated tiers: RAID 5 (3+1) configured on

EFD, RAID 1 on FC, or RAID 6 (6+2) on SATA. Such a policy gives the

greatest flexibility, as at any given time, all the data can be located on the

most appropriate tier.

Note: The default for any newly created policy is to allow 100 percent of the

storage group’s capacity to be placed on any of the included tiers.

In the case of the Custom policy, 10 percent of the configured capacity of

the VP_ProdApp2 storage group can be placed in the EFD tier, 50

percent on the FC tier, and up to 100 percent on both the SATA tier and

external FTS tier.

The No_EFD policy will not move any data to EFD, but does allow 100

percent of the VP_Development storage group’s capacity to be on the

SATA or FTS tier, and up to 50 percent on the FC tier.

If the total for all the tiers combined equals 100 percent and if the thin

devices in the storage group are fully allocated, then each tier is utilized

as set in the policy. For example, if a policy was set up as 10 percent EFD,

20 percent FC, and 70 percent SATA, then 10 percent of the storage

group’s capacity will always be located on the EFD tier, and so on for the

remaining tiers.

FAST policy compliance

A storage group is considered to be compliant with its associated FAST

policy when all data in the storage group is allocated within the bounds

of the upper-usage limits for each tier contained with the policy.

If all of the data in the storage group is allocated within the tiers

contained within the FAST policy, but the allocated capacity in one tier

exceeds the upper-usage limit for that tier, then the storage group is

considered to be noncompliant. In such a case, the FAST controller

attempts to correct this noncompliance by relocating data from the VP

tier where the usage limit is exceeded to one, or both, of the other tiers

contained in the policy. The desired result is to bring the storage group

into compliance.

A special case of noncompliance is when some of the data in the storage

group is allocated in thin pools not contained within any of the VP tiers

Page 14: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

14

FAST VP architecture

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

within the FAST policy. In this case, the storage group is considered to

be noncompliant, and the data considered to be out-of-policy. Again, the

FAST controller attempts to correct this situation by relocating the out-

of-policy data to one, two, or all three of the tiers contained in the policy.

FAST VP architecture

There are two components of FAST VP: Symmetrix Enginuity™ and the

FAST controller.

Symmetrix Enginuity is the storage operating environment that controls

components within the array. The FAST controller is a service that runs

on the service processor.

Figure 3. FAST VP components

Page 15: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

15

FAST VP architecture

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

When FAST VP is active, both components participate in the execution of

two algorithms (the intelligent-tiering algorithm and the allocation-

compliance algorithm) to determine appropriate data placement.

The intelligent-tiering algorithm uses performance data collected by

Enginuity, as well as supporting calculations performed by the FAST

controller, to issue data-movement requests to the VLUN VP data-

movement engine.

The allocation-compliance algorithm enforces the upper limits of storage

capacity that can be used in each tier by a given storage group by also

issuing data-movement requests to the VLUN VP data-movement

engine.

Performance time windows can be defined to specify when the FAST VP

controller should collect performance data. Analysis is then performed to

determine the appropriate tier for devices. By default, performance data

collection occurs 24 hours a day.

Data-movement windows are used to determine when to execute the

data movements necessary to move data between tiers.

Data movements performed by Enginuity are achieved by moving

allocated extents between tiers. The size of data movement can be as

small as 12 tracks, representing a single allocated thin device extent.

More typically, movements are a unit known as an extent group (10 thin

device extents), which is 120 tracks in size.

Note: “FAST VP data movement” on page 34 provides more information on

the actual data movement.

FAST VP has two modes of operation, Automatic or Off. When operating

in the Automatic mode, data analysis and data movements occur

continuously during the defined data movement windows. In the Off

mode, performance statistics continue to be collected, but no data

analysis or data movements take place.

FAST VP state

There are five reported states for FAST VP. These are:

Enabled: All FAST VP functions are performed (performance data

Page 16: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

16

FAST VP configuration

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

collection, performance data analysis, data-movement request generation, and data-movement execution).

Disabled: Only performance data collection is performed. Data analysis is not performed and data movements are not executed.

Disabling: The FAST controller is transitioning from Enabled to Disabled.

DisabledwithError: The FAST controller has stopped operation due to an internal error. Statistics collection and FAST VP performance data movements continue to be performed, however, FAST VP compliance movements are not performed.

Degraded: FAST VP can perform some or all of its functions. However, it cannot perform each function fully.

Note: “Appendix A: FAST VP state” on page 87 provides more information on

each of the FAST controller states.

FAST VP configuration

FAST VP has multiple configuration parameters that control its behavior.

These include time windows that can control when performance data is

collected, included in analysis, and when data movements take place.

Other settings determine the relevance of historical performance data

when analyzed, what percentage of space to reserve in each pool for

activities not associated with FAST VP, and an aggressiveness factor in

generating and executing data-movement requests.

FAST VP time windows

FAST VP utilizes time windows to define certain behaviors regarding

performance data collection and data movement. There are two possible

window types:

Performance time window

Data movement time window

Page 17: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

17

FAST VP configuration

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

The performance time windows are used to specify when performance

data should be collected by Enginuity.

The data movement time windows define when to perform the data

relocations necessary to move data between tiers. Separate data

movement windows can be defined for full LUN movement, performed

by FAST and Optimizer, and sub-LUN data movement, performed by

FAST VP.

Both performance time windows and data movement windows may be

defined as inclusion or exclusion windows. An inclusion time window

indicates that the action should be performed during the defined time

window. An exclusion time window indicates that the action should not

be performed during the defined time window.

There are two methods for defining time windows, legacy and enhanced.

The legacy method uses the Symmetrix Optimizer interface for creating

and managing time windows. The enhanced method uses a time

window management interface.

On Symmetrix VMAX arrays, both the legacy and enhanced methods are

supported. However, only one method can be used at a time. The legacy

method is the default method, but legacy time windows may be

converted to enhanced time windows.

Note: The conversion from legacy to enhanced is a one-way conversion.

There is no method for converting from enhanced to legacy.

Legacy time window characteristics

Using the legacy method, both inclusive and exclusive time windows

may be defined as periodic or non-periodic. Periodic windows allow a

recurrence pattern to be specified as weekly or weekly-by-day. Non-

periodic windows are defined to occur only once. Multiple time

windows, of both performance and data movement, may be defined.

If multiple legacy time windows of the same type have time ranges that

overlap one another, the most recently added time window supersedes

the others, including the system-default time windows.

All defined time windows apply to all devices configured within the

Symmetrix array.

Page 18: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

18

FAST VP configuration

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

A Symmetrix VMAX storage array supports up to 128 defined legacy

time windows. Each time window name may be up to 32 alphanumeric

characters, hyphens (-), and underscores (_).

Enhanced time window characteristics

Using the enhanced method, inclusive windows are defined by

specifying the days of the week and the times in each day during which

performance metrics are collected or data is moved. Each time window is

defined with a start and end time, in 30-minute increments, and the days

of the week to apply the window.

Exclusive windows are defined as a specific time period during which

performance metrics collection or data movement are prevented. Each

exclusive window contains a start date and time, as well as an end date

and time, and may cover several days. The start and end time must be

specified in 30-minute increments.

Multiple inclusive and exclusive enhanced time windows can be defined.

The exclusive time windows have the highest priority and override any

overlapping inclusive time windows.

Enhanced time windows have no associated time-window name.

Performance time window

The performance time windows are used to identify the business cycle

for the Symmetrix array. They specify date and time ranges when

performance samples should be collected, or not collected, for the

purposes of FAST VP performance analysis. The intent of defining

performance time windows is to indicate to FAST which activity periods

should be considered for optimization.

For example, weekends may be used for special utility operations that

should be excluded from optimization. Excluding weekends from the

performance time window means that the activity from the days prior to

the weekend is the best predictor for operations immediately following

the weekend. If the weekend workload is similar to the rest of the week,

Page 19: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

19

FAST VP configuration

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

and the performance sensitivity is the same, then using all days of the

week may be appropriate.

For legacy time windows, a default performance time window includes

all performance data samples, 24 hours a day, 7 days a week, 365 days a

year.

Note: For legacy time windows, in order to prevent FAST VP from collecting

performance statistics on a continuous basis, the first user-defined window

should be created to exclude data collection. Inclusive time windows can

then be created on top of this exclusive time window.

For enhanced time windows, the default behavior is also to collect

performance data, 24 hours a day, 7 days a week. When the first

inclusive performance window is created, the default behavior is

overridden, and performance metrics are only collected during the

defined window.

Data movement time window

Data movement time windows are used to specify date and time ranges

when data movements are allowed, or not allowed. FAST VP data

movements run as low-priority tasks on the Symmetrix back end. They

can introduce additional processing overhead on the back end, however,

host I/O should not be impacted. Data-movement windows can be

planned so they minimize any impact on the performance of other, more

critical workloads.

For legacy time windows, a default data movement time window

prevents any data movement, 24 hours a day, 7 days a week, 365 days a

year.

For enhanced time windows, there is no default data-movement

window. However, until an inclusive window is defined, the default

behavior leads to not allowing any data movement to occur.

FAST VP settings

There are multiple settings that affect the behavior of FAST VP. These

include:

FAST VP Data Movement Mode

Page 20: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

20

FAST VP configuration

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Workload Analysis Period

Initial Analysis Period

Pool Reserved Capacity (PRC)

FAST VP Relocation Rate (FRR)

VP Allocation by FAST Policy

FAST VP Time to Compress

FAST VP Compression Rate

The following sections describe each of these settings, their effect on the

behavior of FAST VP, as well as possible values and the default-setting

values.

FAST VP Data Movement Mode

FAST VP, when enabled, may operate in one of two modes, Automatic

or Off.

In Automatic mode, a data movement request can be generated to move

data based on performance workload. Also, data-movement requests

based on capacity utilization may be generated. These operations are

performed during the periods allowed by the data-movement windows.

In Off mode, no data-movement requests are generated. As a result, no

data movements occur. However, performance metrics continue to be

collected. The default mode is Off.

Workload Analysis Period

The Workload Analysis Period determines the degree to which FAST VP

metrics are influenced by recent host activity, and also less-recent host

activity, that takes place while the performance time window is

considered open.

The longer the time defined in the Workload Analysis Period, the greater

the amount of weight assigned to less-recent host activity.

The Workload Analysis Period can be configured to be between 2 hours

and 4 weeks. The default is 1 week (7 days).

Page 21: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

21

FAST VP configuration

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: For more information on the effect of the Workload Analysis Period,

refer to “FAST VP performance data collection” on page 23.

Initial Analysis Period

The Initial Analysis Period defines the minimum amount of time a thin

device should be under FAST VP management before any performance-

related data movements should be applied. This period only accounts for

time passed while the performance time window is open.

This value should be set to allow sufficient data samples for FAST VP to

establish a good characterization of the typical workload on that device.

This value allows FAST VP to commence analysis and movement

activities on the device, prior to the full Workload Analysis Period

elapsing (if so desired).

The Initial Analysis Period can be configured to be between 2 hours and

4 weeks, however, it cannot exceed the Workload Analysis Period. The

default is 8 hours.

Pool Reserved Capacity

The Pool Reserved Capacity (PRC) reserves a percentage of each pool

included in a VP tier for activities not associated with FAST VP. This

ensures that FAST VP data movements do not fill a thin pool, and

subsequently cause a new extent allocation to fail, as result of a host

write.

When the percentage of unallocated space in a thin pool is equal to the

PRC, FAST VP no longer moves data into that pool. However, data

movements may continue to occur out of the pool to other pools. When

the percentage of unallocated space becomes greater than the PRC, FAST

VP can begin performing data movements into that pool again.

The PRC can be set both system-wide and for each individual pool. By

default, the system-wide setting is applied to all thin pools that have

been included in VP tier definitions. However, this can be overridden for

each individual pool by using the pool-level setting.

The system-wide PRC can be configured to be between 1 percent and 80

percent. The default system-wide PRC is 10 percent.

The pool-level PRC can be configured to be between 1 percent and 80

Page 22: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

22

FAST VP configuration

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

percent, or set to NONE. If the PRC is set to NONE, then the system-

wide setting is used. The default pool-level PRC is NONE.

FAST VP Relocation Rate

The FAST VP Relocation Rate (FRR) is a quality of service (QoS) setting

for FAST VP and affects the aggressiveness of data-movement requests

generated by FAST VP. This aggressiveness is measured as the amount

of data that is requested to be moved at any given time, and the priority

given to moving the data between pools.

The FRR can be configured to be between 1 and 10, with 1 being the most

aggressive. The default is 5.

Note: The rate at which data is moved between pools can also be controlled

by the Symmetrix Quality of Service VLUN setting.

VP Allocation by FAST Policy

The VP Allocation by FAST Policy feature allows new allocations to

come from any of the thin pools included in the FAST VP policy that the

thin device is associated with.

With this feature enabled, FAST VP attempts to allocate new writes in

the most appropriate tier first, based on available performance metrics. If

no performance metrics are available, an attempt is made to allocate to

the pool to which the device is bound.

If the pool initially chosen to allocate the data is full, FAST VP then looks

to other pools contained within the FAST VP policy and allocates from

there. As long as there is space available in at least one pool within the

policy, all new extent allocations will be successful.

The allocation by policy feature is enabled at the Symmetrix array level

and applies to all allocations for all devices managed by FAST VP. The

feature is either Enabled or Disabled. The default setting is Disabled.

When disabled, new allocations only come from the pool to which the

thin device is bound.

Page 23: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

23

FAST VP performance data collection

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: For more information on the decision making process of the VP

allocation by FAST policy feature, refer to “VP Allocation by FAST Policy” on

page 41.

FAST VP Time to Compress

FAST VP Compression automates VP compression at the sub-LUN level

for thin devices that are managed by FAST VP.

The FAST VP Time to Compress parameter enables the use of FAST VP

Compression and sets the inactivity threshold for when data becomes

eligible for compression.

The time to compress can be configured to be between 40 and 400 days,

as well as never. The default is never, meaning FAST VP will not

automatically compress any data.

When set to a particular number of days, any data managed by FAST VP

that has been inactive for at least that number of days will be considered

eligible for compression.

Note: For a thin device’s data to be compressed by FAST VP, the policy the

device is associated with must also contain a pool that has been compression

enabled.

For more information on automated VP compression, refer to “FAST VP

Compression” on page 45.

FAST VP Compression Rate

The FAST Compression Rate is a quality of service (QoS) setting for the

automate VP compression performed by FAST VP. It affects the

aggressiveness with which data is compressed, after it has been inactive

beyond the time to compress value.

The compression rate can be configured to be between 1 and 10, with 1

being the most aggressive. The default is 5.

FAST VP performance data collection

As previously discussed, performance data for use by FAST VP is

collected and maintained by Symmetrix Enginuity. This data is then

Page 24: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

24

FAST VP performance data collection

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

analyzed by the FAST controller and guidelines are generated for the

placement of thin device data on the defined VP tiers within the array.

The following sections discuss the collection and decaying of the thin

device performance metrics.

Thin device performance collection

Performance metrics collected by FAST VP are measured at both the full-

LUN and sub-LUN levels for all thin devices associated with a policy.

At the sub-LUN level, each thin device is broken up into multiple

regions, known as extent groups and extent group sets. Each thin device

is made up of multiple extent group sets which, in turn, contain multiple

extent groups.

Figure 4 shows a graphic representation of a thin device divided into

each of these separate regions.

Figure 4. Thin device performance collection regions

Each extent group is made up of ten contiguous thin device extents. With

each thin device extent being 12 tracks in size, an extent group represents

120 tracks of the device.

Page 25: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

25

FAST VP performance data collection

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Each extent group set is made up of 48 contiguous extent groups,

representing 5,760 tracks of the device.

The metrics collected at each sub-LUN level allow FAST VP to make

separate data-movement requests for each extent group of the device,

120 tracks.

Emulation considerations

FAST VP supports both FBA and CKD emulations. The size of the extent

groups and extent group sets are the same size for both emulations, in

terms of tracks. However, the tracks sizes are different for each of the

emulations, 64 KB for FBA and 57 KB for CKD.

The following table shows the relative size of the thin device extent,

extent group, and extent group set for both FBA and CKD.

Element Tracks FBA CKD

Thin device extent 12 768 KB 684 KB

FAST VP extent

group

120 7,680 KB/

7.5 MB

6,840 KB/

6.7 MB

FAST VP extent

group set

5,760 360 MB 320.6 MB

Performance metrics

When collecting performance data at the LUN level and sub-LUN level

for use by FAST VP, Enginuity only collects statistics related to

Symmetrix back-end activity that is the result of host I/O.

The metrics collected are:

Read miss

Write

Prefetch (sequential read)

The read-miss metric accounts for each DA read operation that is

performed. Reads to areas of a thin device that have not had space

allocated in a thin pool are not counted. Also, read hits, which are

serviced from cache, are not considered.

Write operations are counted in terms of the number of distinct DA

Page 26: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

26

FAST VP performance data collection

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

operations that are performed. The metric accounts for when a write is

destaged. Write hits, to cache, are not considered.

Writes related to specific RAID protection schemes are not counted. In

the case of RAID 1 protected devices, the write I/O is only counted for

one of the mirrors. In the case of RAID 5 and RAID 6 protected devices,

parity writes are not counted.

Prefetch operations are accounted for in terms of the number of distinct

DA operations performed to prefetch data spanning a FAST VP extent.

This metric considers each DA read operation performed as a prefetch

operation.

Workload related to internal copy operations, such as drive rebuilds,

clone operations, VLUN migrations, or even FAST VP data movements,

is not included in the FAST VP metrics.

Note: Performance metrics are only collected during user-defined

performance time windows.

I/O activity rates

For each thin device, a short-term and long-term activity rate is

maintained at the sub-LUN level. Both activity rates are calculated as

weighted moving averages, modeling both recent and less recent

workload on each region of the thin device.

As new performance metrics are collected, the previously stored short-

term and long-term activity rates are decreased by the application of an

exponential decay function. This lessens the influence of previously

collected metrics. The newly collected metrics are then added to the

decayed rates to calculate the new short-term and long-term activity

rates.

The rate of decay applied to the stored I/O rates depends upon the user-

defined Workload Analysis Period, with the short-term rate decayed at a

faster rate than then long-term rate.

Page 27: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

27

FAST VP performance data collection

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: The decay function is only applied to previously collected metrics

while the user-defined performance time window is open.

As an example, when calculating the long-term activity rate with the

default Workload Analysis Period of 168 hours, an I/O that has just been

received is weighted twice as much as an I/O received 24 hours earlier.

Thin device performance score

Based on the performance statistics collected, a prioritized performance

score is generated for the extent groups of each thin device. This score is

calculated by combining two activity rates maintained in Enginuity (the

short-term activity rate and the long-term activity rate) and multiplying

that by a factor determined by the priority assigned to the storage group

when it was associated with the FAST policy.

Then FAST VP places those extent groups with the highest performance

score onto the higher-performing tiers within the associated policy.

Storage group association

The sub-LUN performance metrics previously discussed are collected for

all bound thin devices.

Disassociating a storage group from a FAST policy, or removing devices

from the storage group, causes Enginuity to stop including performance

metrics for those devices in analysis for the purposes of FAST VP.

If the thin device is associated again to a FAST policy, the Initial Analysis

Period must pass for that device before data movements can be

performed once again.

Reassociation

A storage group may be moved from one policy to another, without

interruption to ongoing FAST VP management, by reassociating the

storage group to a new policy.

When a reassociation is performed, all performance metrics previously

collected for the devices in the storage group are maintained within

Enginuity. Also, all attributes of the association, for example priority, are

maintained for the association to the new policy.

Page 28: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

28

FAST VP performance data analysis

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Cache consumption

To maintain the sub-LUN level metrics collected by Enginuity, the

Symmetrix array allocates one cache slot for each thin device that is

bound.

When managing FBA metadevices, cache slots are allocated for both the

metahead and for each of the metamembers.

Note: Each cache slot on a Symmetrix VMAX Series array is one track in size.

FAST VP performance data analysis

FAST VP uses two distinct algorithms, one performance oriented and

one capacity allocation oriented, in order to determine the appropriate

tier to which a device should belong. These algorithms are:

Intelligent-tiering algorithm

Allocation-compliance algorithm

The intelligent-tiering algorithm considers the performance metrics of all

thin devices under FAST VP control and determines the appropriate tier

for each extent group.

The allocation-compliance algorithm is used to enforce the per-tier

storage capacity usage limits.

The following sections provide additional data on each of the algorithms.

Intelligent-tiering algorithm

The goal of the intelligent-tiering algorithm is to use the performance

metrics collected at the sub-LUN level to determine which tier each

extent group should reside in and to submit the needed data movements

to the Virtual LUN (VLUN) VP data-movement engine.

The determination of which extent groups need to be moved is

performed by a task that runs within the Symmetrix array.

Page 29: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

29

FAST VP performance data analysis

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Algorithm structure

The intelligent-tiering algorithm is structured into two components:

A main component that executes within Enginuity

A secondary, supporting, component that executes within the FAST controller on the service processor

The main component assesses whether extent groups need to be moved

to optimize the use of the FAST VP storage tiers. If so, the required data-

movement requests are issued to the VLUN VP data-movement engine.

When determining the appropriate tier for each extent group, the main

component makes use of both the FAST VP metrics, previously

discussed, and supporting calculations performed by the secondary

component on the service processor.

Tier ranking

External tiers can be user-defined as a particular technology, which sets a

performance expectation for that tier.

Within a FAST VP policy, tiers are ranked based on the drive technology

of the pools within the each tier (EFD, FC, or SATA) as well as the

location of the tier (internal or external).

For the six possible tier types that can be included in a FAST VP policy,

the rankings are as follows:

Internal EFD

External EFD

Internal FC

External FC

Internal SATA

External SATA

Note: A FAST VP policy may only contain a maximum of four tiers, up to

three internal and one external.

Promotion-and-demotion thresholds

The supporting component of the intelligent-tiering algorithm, running

on the service processor, does not make explicit data-movement

requests. Rather, it produces a set of promotion-and-demotion

thresholds that define recommendations for data placement across thin

Page 30: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

30

FAST VP performance data analysis

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

tiers.

In a multi-tier configuration, depending on the number of tiers included

in a policy, there are several thresholds that can be generated. These

include:

Internal EFD promotion threshold

External EFD promotion threshold

Internal FC promotion threshold

External FC promotion threshold

Internal SATA promotion threshold

External EFD demotion threshold

Internal FC demotion threshold

External FC demotion threshold

Internal SATA demotion threshold

External SATA demotion threshold

However, as a FAST VP policy can only contain a maximum of four tiers,

each extent group is only compared to up to four thresholds.

Note: There is no internal EFD demotion threshold, as data is never demoted

to the internal EFD tier. Similarly, there is no external SATA promotion

threshold, as data is never promoted to external SATA.

These thresholds allow Enginuity to determine and schedule specific

data movements to optimize the allocation of thin devices under FAST

VP control.

The thresholds are calculated using the performance metrics collected

and the capacity available in each tier. As a result, the thresholds are

dynamic and may increase or decrease as the workload on the devices

managed by FAST VP increases or decreases.

When calculating the thresholds, the goal is to maximize the utilization

of the highest performing tier in the FAST VP policy, while also

proactively demoting inactive or lightly accessed data to the most cost-

effective tier.

Page 31: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

31

FAST VP performance data analysis

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Intelligent-tiering data-movement requests

The size of the data-movement requests generated by the intelligent-

tiering algorithm depends foremost on the amount of capacity that it

deems not to be on the appropriate tier. It also depends on the size of the

VMAX array.

Each back-end disk adapter (DA) in the array is responsible for executing

data movements. The more DAs there are available, the more data that

can be moved at a single time, and, therefore, the larger the request size.

The FAST VP relocation rate does not affect the size of the request

generated, but it does influence the pace at which the requests are

executed and data is moved.

Intelligent–tiering-algorithm execution

Operating within Enginuity, the main component of the intelligent-

tiering algorithm runs continuously during open data-movement

windows, when FAST is enabled and the FAST VP operating mode is

Automatic. As a result, performance-related data movements can occur

continuously during an open data-movement window.

The supporting component, operating within the FAST controller on the

service processor, runs every 10 minutes. It recalculates the promotion-

and-demotion thresholds based on changes in workload on the devices

managed by FAST VP.

Allocation-compliance algorithm

The goal of the allocation-compliance algorithm is to detect and correct

situations where the allocated capacity for a particular storage group

within a thin storage tier exceeds the maximum capacity allowed by the

associated FAST policy.

Policy compliance

A storage group is considered to be in compliance with its associated

FAST policy when the configured capacity of the thin devices in the

storage group is located on tiers defined in the policy, and when the

usage of each tier is within the upper limits of the tier usage limits

specified in the policy.

Compliance violations may occur for multiple reasons, including:

Page 32: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

32

FAST VP performance data analysis

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

New extent allocations performed for thin devices managed by FAST VP

Changes made to the upper-usage limits for a VP tier in a FAST policy

Adding thin devices to a storage group that are themselves out of compliance

Manual VLUN VP migrations of thin devices

Compliance data-movement requests

When a compliance violation exists, the algorithm generates a data-

movement request to return the allocations within the required limits.

This request explicitly indicates which thin device extents should be

moved, and the specific thin pools to which they should be moved.

The size of the data-movement request depends on the amount of

capacity that is currently out of compliance, but also on the user-defined

relocation rate. The maximum size of request that can be generated by

the allocation-compliance algorithm is 10 GB worth of data movements.

When the relocation rate is set to anything other than 1, the FAST

controller divides 10 GB by the relocation rate to determine the new

maximum. For example, if the relocation rate is set to 2, the maximum

request size is 5 GB; if it is 10, the maximum size is 1 GB.

Note: The allocation-compliance algorithm is not designed to be a tool for

bulk data movement. To move large amounts of data from one pool to

another, for a given set of devices, VLUN VP should be used. VLUN VP

allows both a source pool and a target pool to be specified to zero in on the

exact data to be migrated.

Intelligent-tiering-algorithm coordination

The compliance algorithm attempts to minimize the amount of

movements performed to correct compliance that may, in turn, generate

movements performed by the intelligent-tiering algorithm.

Page 33: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

33

FAST VP performance data analysis

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Compliance violations can be corrected by coordinating the movement

requests with the analysis performed by the intelligent-tiering algorithm

to determine the most appropriate extents to be moved to the most

appropriate tier.

Allocation-compliance-algorithm execution

The compliance algorithm runs every 10 minutes during open data-

movement windows, when FAST is enabled, and the FAST VP operating

mode is Automatic.

FAST controller availability

Should the FAST controller, a resident on the service processor, become

unavailable, FAST VP will continue operating, albeit in a degraded

mode. During this time, each component of FAST VP is affected in a

different way.

In the case of the intelligent-tiering algorithm, the algorithm is structured

into two components:

The main component, which executes within Enginuity

A supporting component that executes within the FAST controller (on the SP)

The FAST controller is responsible for performing supporting analysis,

calculating the promotion-and-demotion thresholds. Each time the

thresholds are recalculated and passed to Enginuity, an expiration

timestamp is placed on the thresholds and is set at four hours from

creation. At the same time, a more conservative, backup set of thresholds

is created with an expiration timestamp of four days from creation.

In the case where the FAST controller becomes unavailable and no new

thresholds are generated, Enginuity will use the last good thresholds for

four hours, and then switch to the backup thresholds for four days. After

four days, the performance movements will stop.

Performance data collection is unaffected by a controller outage, as it

takes place completely within Enginuity. The short-term and long-term

activity rates are also unaffected, as they are calculated and stored within

Enginuity. No performance data information is lost.

When the FAST controller is restarted, the most recent activity rates are

read from Enginuity, and new thresholds are calculated based on the

Page 34: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

34

FAST VP data movement

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

most recent workload information.

In the case of the allocation-compliance algorithm, however, when the

FAST controller becomes unavailable, no compliance-based movement

requests are generated. As such, the ability of FAST VP to correct

allocation-compliance violations is impaired. However, performance

movements may correct compliance as it moves data.

FAST VP data movement

There are two types of data movement that can occur under FAST VP:

Intelligent-tiering-algorithm-related movements and allocation-

compliance-algorithm-related movements. Both types of data movement

occur only during user-defined data movement windows.

Intelligent-tiering-algorithm-related movements are requested and

executed by Enginuity. These data movements are governed by the

workload on each extent group, but are executed only within the

constraints of the associated FAST policy. This is to say that performance

movements will not cause a storage group to become noncompliant with

its FAST policy.

Allocation–compliance-algorithm-related movements are generated by

the FAST controller, and executed by Enginuity. These movements bring

the capacity of the storage group back within the boundaries specified by

the associated policy. Performance information from the intelligent-

tiering algorithm is used to determine the more appropriate sub-extents

to move when restoring compliance.

Data-movement engine

Data movements executed by FAST VP are performed by the VLUN VP

data-movement engine. They involve moving thin device extents

between thin pools within the array.

Page 35: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

35

FAST VP data movement

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Extents are moved by way of a move process only. Extents are not

swapped between pools.

The movement of extents, or extent groups, does not change the thin

device binding information. The thin device remains bound to the pool it

was originally bound to. New allocations for the thin device, as the result

of host writes, continue to come from the bound pool, unless VP

allocation by FAST VP is enabled.

Movement considerations

To complete a move, the following must hold true:

The FAST VP operating mode must be Automatic.

The VP data movement window must be open.

The thin device affected must not be pinned.

There must be sufficient unallocated space in the thin pools included in the destination tier to accommodate the data being moved.

The destination tier must contain at least one thin pool that has not exceeded the pool reserved capacity (PRC).

Note: If the selected destination tier contains only pools that have reached the

PRC limit, then an alternative tier may be considered by the movement task.

Other movement considerations include:

Only extents that are allocated are moved.

No back-end configuration changes are performed during a FAST VP data movement, and, as a result, no configuration locks are held during the process.

As swaps are not performed, there is no requirement for any swap space, such as DRVs, to facilitate data movement.

Data-movement process

The following section details the process followed during a FAST VP

data movement.

In the following illustration, device 100 is bound to an FC thin pool. The

device is associated with a FAST policy that also contains tiers with an

EFD pool and a SATA pool.

Over time, FAST VP has determined that several of the devices extents

Page 36: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

36

FAST VP data movement

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

need to be moved to the other tiers within the policy.

As data movements can occur continuously during the thin data

movement time window, a single device’s allocations may change

multiple times while it is actively managed by FAST VP.

Initial allocation

Figure 5 shows the thin device with its initial binding and allocation on a

FC tier. The additional tiers the device is associated with as part of the

FAST policy are also shown.

Figure 5. FAST VP data movement: Initial allocation and associated tiers

Extent movements

After the initial analysis period has passed, the performance metrics

collected for device 100 are analyzed by the FAST controller. Based on

this performance analysis, it is determined that several of the allocated

extents need to be moved to the other tiers in the FAST policy.

Page 37: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

37

FAST VP data movement

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Due to a higher level of activity, the extents labeled 1 and 6 are promoted

to the EFD tier. Meanwhile, the extents labeled 2, 3, 7, and 8 are demoted

to SATA, as they were determined to be less active.

The data movements are queued up on each DA, and the data is

transferred. When the data transfer is complete, the space originally

consumed by the extents in the FC tier is deallocated and is reported as

free space in that tier, as shown in Figure 6.

Figure 6. FAST VP data movement: Extent relocations

Even though data has been moved to other tiers within the array, the

thin device remains bound to the pool it was originally bound to, which

is contained in the FC tier.

New host writes

By default, new allocations that occur as a result of host writes come

from the pool to which the thin device is bound.

Page 38: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

38

FAST VP data movement

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: The default behavior of allocating new extents from the pool to which

the device is bound may be changed. See “VP Allocation by FAST Policy” on

page 41.

Figure 7 shows additional extents 9 and 10 that have been allocated in

the pool.

Figure 7. FAST VP data movements: New host writes

Continuous operations

Over longer periods of time, new data is generated, causing more

allocations. Also, data access patterns may change, causing additional

promotions and demotions to be performed.

Figure 8 shows a snapshot of device 100 and its data allocation across all

three tiers.

Page 39: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

39

FAST VP data movement

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Figure 8. FAST VP data movement: Continuous operations

Data movement control

Several mechanisms exist for controlling the movement of data related to

thin devices under FAST VP control.

Device pinning

To prevent FAST VP changing the current tiering allocation of a thin

device, a feature called device pinning may be used. Pinning a device

locks all extent allocations for the device in their current locations, and

prevents FAST VP from relocating them.

Any new allocations performed for a pinned device comes from the thin

pool to which the device is bound. If allocation by policy is enabled, new

allocations for a pinned thin device will be performed as if no

performance metrics are available for the device. These allocations are

not moved by FAST VP.

The thin device needs to be unpinned to re-enable data movement.

Page 40: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

40

FAST VP data movement

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: While a device is pinned, performance statistics continue to be

collected for that device.

FAST policy disassociation

Aside from pinning a device to prevent a thin device’s allocation from

being changed by FAST VP, the device may be removed from the storage

group that is associated to a FAST policy. Removing the device from the

storage group disassociates it from the FAST policy and removes the

device from the control of FAST VP. As a result, no further data

movements are performed for that device. Allocated data previously

moved by FAST VP remains in its current locations and is not

automatically returned to the thin device’s bound pool.

Note: Disassociating a device from a FAST policy prevents performance

statistics from being collected for that device, and all previously collected

metrics are discarded. If the thin device is associated again to a FAST policy,

the Initial Analysis Period must pass for that device before data movements

can once again be performed.

Changing the quality of service

If the performed data movements by FAST VP are impacting other

applications or replication tasks within the Symmetrix array, quality of

service (QoS) tools may be used to change the pace at which data is

moved. Slowing down the FAST VP data movements gives higher

priority to the other tasks running on the Symmetrix back end.

The VLUN copy pace can be set between 0 and 16 inclusively, with 0

being the fastest and 16 being the slowest. The default QoS value is 0.

Setting the pace value to 16 gives the FAST VP data movements the

lowest priority on the array. Once the unrelated copy tasks have

completed, the pace setting can be reset to 0.

Page 41: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

41

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: QoS can be used to change the copy pace for individual devices under

FAST VP control. The FAST Relocation Rate can be used to change the copy

pace for all devices under FAST VP control.

Operating mode

To stop data movements for all thin devices under FAST VP control, the

operating mode may be set to Off. While off, performance statistics

continue to be collected by Enginuity. However, no data-movement

requests are generated by Enginuity or the FAST controller.

Advanced FAST VP features

Along with the features already described, advanced features exist that

further improve the ease of management and usability of FAST VP.

These advanced features include:

VP Allocation by FAST Policy: Allows thin device allocations to come from thin pools other than the pool to which the devices are bound.

FAST VP SRDF® coordination: Allows promotion-and-demotion decisions of data belonging to a thin R2 device to account for workload on the corresponding thin R1 device.

FAST VP Compression: Automates VP compression at the sub-LUN level for thin devices that are managed by FAST VP.

User-defined FTS tier: Allows an external FTS tier to be specified as something other than the lowest tier.

The following sections describe each of these advanced features in detail.

VP Allocation by FAST Policy

By default, new extent allocations generated by writes to a thin device

come from the thin pool to which the device is bound. This behavior has

two potential consequences. First, data allocated from the bound pool

may be promoted or demoted shortly after the allocation that causes an

additional movement of data on the back end. Second, should the bound

pool fill up, a write generating a new allocation to the pool may fail.

The VP Allocation by FAST Policy feature addresses these potential

consequences for devices managed by FAST VP. When the feature is

Page 42: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

42

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

enabled, new allocations can come from any of the thin pools included in

the FAST VP policy to which the thin device is associated. (Allocations

are not limited to just the bound pool.)

The allocation by policy feature is a setting that is either Enabled or

Disabled across the entire Symmetrix array. The default setting is

Disabled. When disabled, new allocations come only from the pool to

which the thin device is bound.

When this feature is enabled, FAST VP attempts, based on available

performance metrics, to direct the allocation to the most appropriate tier.

If the selected pool is full, an alternate pool is chosen from the other thin

pools included within the FAST VP policy.

The following section describes the decision-making criteria for selecting

the thin pool from which to allocate.

Allocation by policy decision criteria

As previously mentioned, FAST VP collects performance metrics at three

distinct levels for a thin device: Full LUN, extent group set, and extent

group. Performance metrics already collected for each extent group set

can be applied to newly allocated data within each of those sets.

When allocation by policy is enabled on a Symmetrix array, FAST VP

first checks for the existence of performance metrics on the extent group

set for which the new allocation is generated.

If available, the performance metrics are used to determine the most

appropriate tier, from a workload perspective, from which to allocate the

extent. Once the tier has been selected, a pool within the tier is selected,

and the allocation is attempted.

If there are no performance metrics available for the extent group set

being written to (this would happen when the entire set is unallocated),

FAST VP attempts to allocate new extents from the pool to which the

device is bound.

Page 43: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

43

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

In either of these cases, should the selected pool be full, an alternate pool

within the policy is chosen.

To determine the alternate pool, the tiers are sorted in ascending order of

the configured capacity of each tier. Each tier is then checked for

available space within each pool in the tier, and the allocation attempted.

Should the selected pool be full, FAST VP moves onto the next pool

within the tier (if it exists), or the next tier on the list.

If a selected tier contains multiple pools, the allocation pool is selected at

random.

This continues until the allocation is successful, or until a determination

is made that all pools within the policy are full. As long as there is space

available in at least one of the pools within the policy, all new extent

allocations will be successful.

Note: If the device generating the allocation is pinned, the existence of

performance metrics is ignored. FAST VP first attempts to allocate from the

bound pool. If the bound pool is full, an alternate pool within the policy is

chosen.

FAST VP SRDF coordination

FAST VP has no restrictions in its ability to manage SRDF devices.

However, it must be considered that FAST VP data movements are

restricted to the array upon which FAST VP is operating. By default,

however, there is no coordination of data movements between the source

and target arrays. FAST VP acts independently on both.

While an R1 device typically undergoes a read-and-write workload mix,

the corresponding R2 device only sees a write workload. (Reads against

the R1 are not propagated across the link.) A consequence of this is that

the R2 device data may not be located on the same tier as the related data

on the R1 device.

FAST VP SRDF coordination allows R1 performance metrics to be used

when making promotion-and-demotion decisions for data belonging to

an R2 device.

SRDF coordination is Enabled or Disabled by the storage group

associated with a FAST VP policy. The default state is Disabled.

Page 44: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

44

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: FAST VP SRDF coordination requires that both the local and remote

Symmetrix array be at a code level that supports SRDF coordination.

This feature can be enabled when the storage group is associated with a

policy. It can also be enabled by modifying an existing association.

You only need to enable this feature on a storage group containing R1

devices. If it is enabled on a storage group containing R2 devices, the

setting has no effect. The only case that this would not be true is if an

SRDF swap operation occurs and converts the R2 devices to R1 devices.

FAST VP SRDF coordination is supported for single, concurrent, and

casdcaded SRDF pairings in any mode of operation: Synchronous,

asynchronous, adaptive copy, SRDF/EDP, and SRDF/Star.

In the case of cascaded SRDF configurations, the R21 device acts as a

relay, transmitting the R1 metrics to the R2 device.

In the case of an R22 device, metrics will only be received across the

SRDF link that is currently active.

Note: FAST VP SRDF coordination is supported in mainframe environments

running PPRC.

Effective performance score

When enabled on a storage group, the collected FAST VP performance

metrics for R1 devices are periodically transmitted across the SRDF link

to the corresponding R2 devices.

Note: Performance metrics are only sent for R1 devices when their

corresponding R2 devices are being managed by FAST VP.

On the R2 device, the R1 performance metrics are merged with the actual

R2 metrics. This creates an effective performance score for data on the R2

devices.

Page 45: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

45

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

When promotion-and-demotion decisions are made on the remote array,

the effective score is used for the R2 data, thereby allowing the R1

workload to influence the movement of the R2 data. Data that is heavily

read on the R1 device is likely to be promoted to the higher tiers in the

policy to which the R2 device is associated.

Note: The R2 device’s effective performance score also influences the

calculation of promotion-and-demotion thresholds on the remote array.

SRDF coordination considerations

When the SRDF link between the R1 and R2 devices is not ready, the R1

performance metrics are not transmitted to the R2 device. When the link

is restored, the metrics are transmitted again.

The SRDF link is considered to be not ready when the SRDF pair state is

in one of the following states:

Split

Suspended

Failedover

Partitioned

During the period that the metrics are not being sent, previously

received R1 metrics are decayed in a manner similar to normally

collected performance data.

Note: For more information on decaying of performance metrics, see “I/O

activity rates” on page 26.

If an SRDF personality swap operation is performed, performance

metrics are only transmitted when SRDF coordination is enabled on the

storage group containing the new R1 devices.

Note: After a swap operation, performance metrics are automatically

transmitted from the new R1 devices, as long as SRDF coordination had been

previously enabled while the storage group contained R2 devices.

FAST VP Compression

VP Compression provides the ability to compress data belonging to thin

devices. Compressed data consumes less physical space in a thin pool,

Page 46: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

46

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

thus providing additional space for more data to be written to the pool.

EMC compression capabilities for Virtually Provisioned and FAST VP

environments provide customers with 2:1 or greater data compression

for lightly and very infrequently accessed data. EMC defines

infrequently accessed data as that which is not accessed within month-

end, quarter-end, or year-end activities, nor accessed as part of full

backup processing.

Data can be compressed manually for an individual thin device, or

group of devices, by means of Solutions Enabler and Unisphere for

VMAX.

Alternatively, inactive data belonging to thin devices managed by FAST

VP can be automatically compressed using FAST VP Compression.

Note: For more information on VP Compression, refer to the Best Practices

for Fast, Simple Capacity Allocation with EMC Symmetrix Virtual

Provisioning Technical Note, available at http://support.emc.com.

FAST VP Compression is enabled at the system level by setting the time

to compress parameter to a value between 40 and 400 days. Data that is

seen to be inactive for a period of time longer than this parameter will be

considered eligible for automatic compression. The default value for the

time to compress is never.

Automated compression

FAST VP Compression automates VP Compression for thin devices that

are managed by FAST VP. This automated compression takes place at

the sub-LUN level. Compression decisions are based on the activity level

of individual extent groups (120 tracks), similar to promotion-and-

demotion decisions. However, the actual compression itself is performed

per thin device extent (12 tracks).

Extent groups that have been inactive on disk for a user-defined period

of time, known as the time to compress, will be considered eligible for

Page 47: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

47

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

compression. The time to compress parameter can be set between 40 and

400 days. The default time to compress is never.

In order for thin device extent to be compressed automatically by FAST

VP, all of the following criteria must be met:

The extent must belong to an extent group that has been inactive for a period equal to or greater than the time to compress.

The extent must be located in a thin pool that has been enabled for compression.

The extent must compress by at least 50 percent.

Policy compliance

For the purposes of calculating a storage group’s compliance with a

policy, FAST VP does not take into account any compression that has

occurred on any of the devices within the group. Only the logical,

allocated capacity consumed by the devices is considered.

For example, if 50 GB of data belonging to a storage group was demoted

to the SATA tier and subsequently compressed to only consume 20 GB,

the compliance calculation for the storage group would be based on 50

GB consumed within the SATA tier.

Pool reserved capacity

The pool reserved capacity (PRC) reserves a percentage of each pool

included in a VP tier for activities not associated with FAST VP. The PRC

is based on the actual capacity allocated within a thin pool. This applies

even if there is compressed data allocated in the pool.

Compression task execution

Extent groups are evaluated for compression at the same time extent

groups are evaluated for promotion or demotion. However,

performance-related promotions and demotions have a higher priority

than compression activity, so there may be some time lag between an

extent group becoming eligible for compression and it actually being

compressed.

SRDF interoperability

By default, FAST VP runs independently on each side of an SRDF link,

and movement decisions are based on the workload seen by the data

locally on each array. This applies to FAST VP Compression, too. This

Page 48: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

48

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

means that compression decisions made on the R1 side may not

necessarily be made on the R2 side.

If FAST VP SRDF coordination is enabled for a storage group associated

with a FAST VP policy, performance metrics collected for an R1 device

are transferred to the R2, allowing the R1 workload to influence

promotion-and-demotion decisions on the R2 data.

As a part of the performance metrics collected, the length of time the

data has been inactive is also transferred. This period of inactivity on R1

data can then influence compression decisions on the corresponding R2

data.

This does not guarantee, however, that data compressed on the R1 side

will have its corresponding data compressed on the R2 side, and vice

versa.

Compression decisions on each side of the SRDF link will be influenced

by the Time to Compress parameter configured on each array, and also

by the pools that are enabled for compression.

User-defined FTS tier

By default, any external tier added to a FAST VP policy is considered to

be the lowest performing tier within the policy, even below the SATA

tier. However, depending on the configuration of the external tier, it may

be possible that the tier can provide a higher level of performance than

an internal SATA, or even FC, tier.

In the case where it is expected that an external tier will perform at a

higher level than one or more internal tiers, the relative performance

expectation can be user-defined. This will then affect the tiers ranking

amongst other tiers when added to a FAST VP policy.

Page 49: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

49

Advanced FAST VP features

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: Performance of external arrays may vary under changing workloads,

especially when the external array is not fully devoted to VMAX I/O. As a

result, FAST VP may not always be able to make the optimal choices for data

placement when used with FTS-based tiers defined as other than the lowest

tier of storage.

External technology type

When an external tier is created, an optional technology type (EFD, FC,

or SATA) may be specified, in addition to the external location. By

specifying a technology type, a related expectation of performance is

associated with the tier.

For example, an external tier created with the technology type of FC will

be expected to have similar performance characteristics to that of an

internal FC tier, with some allowance for latency that may be introduced

by the SAN connectivity between the local and external storage arrays.

As such, an external FC tier will be ranked below an internal FC tier, but

above an internal SATA tier when included in a multi-tier policy.

After an external tier has been created, the technology type can be

modified. If the type is changed, the ranking of the tier will change

within any policy in which it has been included. As such, an external tier

can be upgraded or downgraded within a policy.

Note: The technology type of a tier can only be modified for an external tier.

It cannot be changed for an internal tier.

External tier performance discovery

When a technology type, other than SATA, is specified for an external

tier, the tier is potentially expected to perform at a higher level than

other tiers contained within a FAST VP policy. However, there may be

circumstances where this is not the case, and the performance

expectations cannot be met.

For example, unexpected latency in the SAN may cause the performance

of an external tier to fall below that of an internal tier of a lower-ranked

technology. Having a higher-ranked tier perform less than a lower-

ranked tier could potentially have an adverse affect on the behavior of

FAST VP.

Page 50: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

50

FAST VP interoperability

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

To address this possibility, Enginuity executes an initial performance

discovery of the tier when it is first added to a FAST VP policy. This is

done to ensure that the performance of the external tier is in line with the

user-defined expectations. This initial performance test is known as full

discovery testing. Similar testing is performed when a change is made to

the underlying eDisks for the tier, or if the technology type is modified in

the tier definition.

Once the full performance discovery completes, an ongoing task

periodically executes a lighter performance discovery. This is known as

discovery validation testing. The purpose of this discovery is to validate, or

incrementally adjust, the previously discovered performance.

At any point, if the performance of an external tier falls below

expectations, an event is triggered, alerting users to this fact.

The event (event ID 1511) is triggered when the performance of an

external tier changes, dropping below the expected level.

Based on this alert, users can address the discrepancy between the actual

performance and the expected performance of the tier. This can be done

by either addressing the cause of the degraded performance on the

external tier, or by lowering the expectations of the tier.

FAST VP interoperability

FAST VP is fully interoperable with all Symmetrix replication

technologies: EMC SRDF, EMC TimeFinder®/Clone, TimeFinder/Snap,

and Open Replicator. Any active replication on a Symmetrix device

remains intact while data from that device is being moved. Similarly, all

incremental relationships are maintained for the moved or swapped

devices.

FAST VP also operates alongside Symmetrix features such as Symmetrix

Optimizer, Dynamic Cache Partitioning, and Auto-provisioning Groups.

Page 51: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

51

FAST VP interoperability

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

SRDF

Thin SRDF devices, R1 or R2, can be associated with a FAST policy.

Extents of SRDF devices can be moved between tiers while the devices

are being actively replicated, in either synchronous or asynchronous

modes.

Note: For more information on using FAST VP with SRDF, see “FAST VP

SRDF coordination” on page 43.

TimeFinder/Clone

Both the source and target devices of a TimeFinder/Clone session can be

managed by FAST VP. However, the source and target are managed

independently, and, as a result, may end up with different extent

allocations across tiers.

TimeFinder/Snap

The source device in a TimeFinder/Snap session can be managed by

FAST VP. However, the target device (VDEV) may not be brought under

FAST VP control.

TimeFinder VP Snap

The source device in a TimeFinder VP Snap session can be managed by

FAST VP. Target devices may also be managed by FAST VP, however,

extent allocations that are shared by multiple target devices are not

moved.

Open Replicator

The control device in an Open Replicator session, push or pull, can have

extents moved by FAST VP.

Virtual Provisioning

Each thin device, whether under FAST VP control or not, may only be

bound to a single thin pool. All host-write-generated allocations, or user-

requested pre-allocations, are performed on this pool. FAST VP data

movements do not change the binding information for a thin device.

It is possible to change the binding information for a thin device without

changing any of the current extent allocations for the device. However,

when rebinding a device that is under FAST VP control, the thin pool the

device is being rebound to must belong to one of the VP tiers contained

Page 52: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

52

FAST VP interoperability

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

in the policy with which the device is associated.

Virtual Provisioning space reclamation

Space reclamation may be run against a thin device under FAST VP

control. However, during the space-reclamation process, the sub-LUN

performance metrics are not updated, and no data movements are

performed.

Note: If FAST VP is actively moving extents of a device, a request to reclaim

space on that device will fail. Prior to issuing the space-reclamation task, the

device should first be pinned. Pinning a device suspends any active FAST VP

data movements for the device and allows the request to succeed.

Virtual Provisioning T10 unmap

Unmap commands can be issued to thin devices under FAST VP control.

The T10 SCSI unmap command for thin devices advises a target thin

device that a range of blocks are no longer in use. If this range covers a

full thin device extent, then that extent can be deallocated, and the free

space is returned to the pool.

If the unmap command range covers only some tracks in an extent, those

tracks are marked Never Written by Host (NWBH). The extent is not

deallocated. However, those tracks do not have to be retrieved from the

drive should a read request be performed. Instead, the Symmetrix array

immediately returns all zeros.

Note: The T10 SCSI unmap command is only supported in Open Systems

(FBA) environments.

Virtual Provisioning pool management

Data devices may be added to or removed from a thin pool that is in the

FAST VP tier. Data movements related to FAST VP, into or out of the

thin pool, continue while the data devices are being modified.

Page 53: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

53

FAST VP interoperability

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

When adding data devices to a thin pool, pool rebalancing may be run.

Similarly, when disabling and removing data devices from the pool, they

drain their allocated tracks to other enabled data devices in the pool.

While running either data-device draining or automated pool

rebalancing on a thin pool that is included in a VP tier is allowed, both

processes may affect performance of FAST VP data movements.

Virtual LUN VP mobility

A thin device under FAST VP control may be migrated using VLUN VP.

Such a migration results in all allocated extents of the device being

moved to a single thin pool.

While the migration is in progress, no FAST VP related data movements

are performed. Once the migration is complete, all allocated extents of

the thin device will be available to be retiered.

To prevent the migrated device from being retiered by FAST VP

immediately following the migration, it is recommended that the device

first be pinned. The device can later be unpinned to re-enable FAST VP-

related data movements.

FAST

Both FAST and FAST VP may coexist within a single Symmetrix array.

FAST only performs full-device movements of non-thin devices. As a

result, there is no impact to FAST VP’s management of thin devices.

Both FAST and FAST VP share some configuration parameters. These

are:

Workload Analysis Period

Initial Analysis Period

Performance Time Windows

Symmetrix Optimizer

Symmetrix Optimizer operates only on non-thin devices. As such, there

is no impact on FAST VPs management of thin devices.

Both Optimizer and FAST VP share some configuration parameters.

These are:

Workload Analysis Period

Initial Analysis Period

Page 54: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

54

FAST VP interoperability

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Performance Time Windows

Dynamic Cache Partitioning

Dynamic Cache Partitioning (DCP) can be used to isolate storage

handling of different applications. As data movements use the same

cache partition as the application, movements of data on behalf of one

application do not affect the performance of applications that are not

sharing the same cache partition.

FBA Auto-provisioning Groups

Storage groups created for the purposes of Auto-provisioning FBA

devices may also be used for FAST VP. However, while a device may be

contained in multiple storage groups for the purposes of Auto-

provisioning, the device may only be contained in one storage group that

is associated with a FAST policy (DP or VP).

If a storage group contains a mix of device types, thin and non-thin, only

the devices matching the type of FAST policy the group is associated

with are managed by FAST.

If it is intended that both device types in an Auto-provisioning storage

group be managed by FAST and FAST VP, then separate storage groups

need to be created. A storage group with the non-thin devices may then

be associated with a policy containing DP tiers. A separate storage group

containing the thin devices needs to be associated with a policy

containing VP tiers.

If separate storage groups are created for the purposes of applying

separate FAST policies, then these groups can be added to a parent

storage group, using the cascaded SG feature. A masking view can then

be applied to the parent SG, provisioning both sets of devices.

Note: FAST policies may only be associated to storage groups containing

devices. A parent SG, containing other storage groups, cannot be associated

to a FAST policy.

Page 55: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

55

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Best practice considerations and recommendations

The following sections detail best practice recommendations for

planning the implementation of a FAST VP environment.

The best practices documented are based on features available in

Enginuity 5876.229.145, Solutions Enabler V7.6, and Unisphere for

VMAX 1.6.

FAST VP configuration parameters

FAST VP has multiple configuration parameters that control its behavior.

These include settings to determine the effect of past workloads on data

analysis, quality of service for data movements, and pool space to be

reserved for activities not associated with FAST VP. Also, performance-

collection and data-movement time windows can be defined.

The following sections describe best practice recommendations for each

of these configuration parameters.

Note: For more information on each of these configuration parameters, refer

to the Implementing Fully Automated Storage Tiering for Virtual Pools

(FAST VP) for EMC Symmetrix VMAX Series Arrays Technical Note

available at http://support.emc.com.

Performance time window

The performance time windows specify date and time ranges what

collected performance metrics should be included, or not, for the

purposes of FAST VP performance analysis. By default, performance

metrics are collected 24 hours a day, every day. Time windows may be

defined, however, to include only certain times of day or days of the

week, as well as exclude other time periods.

While the performance time window is open, in addition to performance

metrics being collected, the long-term and short-term activity rates are

also being incremented and decayed. As such, the performance time

window can greatly influence the performance scores of data.

For example, if some data is only busy during production hours of 8 a.m.

to 8 p.m., limiting the performance time window to just these hours can

Page 56: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

56

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

cause the data accessed during production hours to much higher scores

than the default 24-hour window.

As a best practice, the performance time windows should be configured

to target the specific workloads the user wishes to optimize. For

example, if the normal production workload typically occurs between 8

a.m. and 8 p.m. on weekdays then the performance time windows

should be configured to match these periods.

Similarly, if the workload on weekends or during holiday periods, are

not as important or representative of the other days; you should consider

excluding them from the performance time window.

Note: The performance time window is applied system-wide. If multiple

applications are active on the array, but active at different times, including

weekends, then the default performance time window behavior should be

left unchanged.

Data movement time window

Data movement time windows are used to specify date and time ranges

when data movements are allowed, or not allowed, to be performed.

The best practice recommendation is that, at a minimum, the data

movement window should allow data movements for the same period of

time that the performance time windows allow data collection. This

allows FAST VP to react quickly and dynamically to any changes in

workload that occur on the array.

Unless there are specific time periods to avoid data movements, during a

backup window, for example, it may be appropriate to set the data

movement window to allow FAST VP to perform movements 24 hours a

day, every day.

Note: If there is a concern about possible impact of data movements

occurring during a production workload, the FAST VP Relocation Rate can

be used to minimize this impact.

Page 57: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

57

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Workload Analysis Period

The Workload Analysis Period (WAP) determines the degree to which

FAST VP metrics are influenced by recent host activity, and also less-

recent host activity, that takes place while the performance time window

is considered open.

The longer the time defined in the WAP, the greater the amount of

weight is assigned to less-recent host activity.

Lowering the WAP may allow FAST VP to react to changes in workload

more quickly, but may lead to greater amounts of data being moved

between tiers.

The influence of the WAP setting is applied only when the performance

time window is open. For example, if the performance time window is

left at the default (open 24 hours a day, 7 days a week), the default WAP

(one week) will cover one calendar week. However, if the performance

window is configured to be open 12 hours a day, 7 days a week, the

default WAP will actually incorporate two calendar weeks.

The best practice recommendation for the Workload Analysis Period is

to match it to the performance time window to cover one calendar week.

Initial Analysis Period

The Initial Analysis Period (IAP) defines the minimum amount of time a

thin device should be under FAST VP management before any

performance-related data movements should be applied.

This parameter should be set to a value that allows sufficient data

samples for FAST VP to establish a good characterization of the typical

workload on that device.

At the initial deployment of FAST VP, it may make sense to set the IAP

to 168 hours (one week) to ensure that a typical weekly workload cycle is

seen. However, once FAST VP data movement has begun, you may

reduce the IAP to 24 hours (one day), which allows the newly associated

devices to benefit from FAST VP movement recommendations more

quickly.

FAST VP Relocation Rate

The FAST VP Relocation Rate (FRR) is a quality of service (QoS) setting

for FAST VP. The FRR affects the aggressiveness of data-movement

Page 58: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

58

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

requests generated by FAST VP. This aggressiveness is measured as the

amount of data that is requested to be moved at any given time, and the

priority given to moving the data between pools.

Setting the FRR to the most aggressive value, 1, causes FAST VP to

attempt to move the most data it can, as quickly as it can. Depending on

the amount of data to be moved, an FRR of 1 is more likely to impact

host I/O response times, due to the additional back-end overhead being

generated by the FAST VP data movements. However, the distribution

of data across tiers completes in a shorter period of time.

Setting the FRR to the least aggressive value, 10, causes FAST VP to

greatly reduce the amount of data that is moved, and the pace at which it

moves. This setting causes no impact to host I/O response time, but the

final redistribution of data across tiers takes longer.

Figure 1 shows the same workload, 1,500 IOPS of type OLTP2, being run

on an environment containing two FAST VP tiers, Fibre Channel (FC)

and Enterprise Flash (EFD). The same test was carried out with three

separate relocation rates: 1, 5, and 8.

With a FRR of 1, an initial increase in response time is seen at the two-

hour mark, when FAST VP data movement began, while no increase in

response time is seen when the relocation rate is set to 8. However, the

steady state for the response time is seen in a much shorter period of

time for the lower, more aggressive, setting.

Page 59: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

59

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Figure 9. Example workload with varying relocation rates

The default value for the FRR is 5. However, the best practice

recommendation for the initial deployment of FAST VP is to start with a

more conservative value for the relocation rate, perhaps 7 or 8. The

reason for this is that when FAST VP is first enabled, the amount of data

to be moved is likely to be greater, compared to when FAST VP has been

running for some time.

At a later date, when it is seen that the amount of data movements

between tiers is less, the FRR can be set to a more aggressive level,

possibly 2 or 3. This allows FAST VP to adjust to small changes in the

workload more quickly.

Pool Reserved Capacity

The Pool Reserved Capacity (PRC) reserves a percentage of each pool

included in a VP tier for activities not associated with FAST VP. When

the percentage of unallocated space in a thin pool is equal to the PRC,

FAST VP no longer performs data movements into that pool.

The PRC can be set both as a system-wide setting and for each individual

pool. If the PRC has not been set for a pool, or the PRC for the pool has

Page 60: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

60

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

been set to NONE, then the system-wide setting is used.

For the system-wide setting, the best practice recommendation is to use

the default value of 10 percent.

For individual pools, if thin devices are bound to the pool, the best

practice recommendation is to set the PRC based on the lowest allocation

warning level for that thin pool. For example, if a warning is triggered

when a thin pool has reached an allocation of 80 percent of its capacity,

then the PRC should be set to 20 percent. This ensures that the remaining

20 percent of the pool is only used for new host-generated allocations,

and not FAST VP data movements.

If no thin devices are bound to the pool, or are going to be bound, then

the PRC should be set to the lowest possible value, 1 percent.

Note: In an environment where only two tiers are configured in the VMAX

storage array, and VP allocation by FAST policy is enabled, it may be

beneficial to set the PRC to 1 percent for all thin pools.

VP Allocation by FAST Policy

The VP Allocation by Fast Policy feature allows new allocations to come

from any of the thin pools included in the FAST VP policy to which the

thin device is associated.

With this feature enabled, FAST VP attempts to allocate new writes in

the most appropriate tier first, based on available performance metrics. If

no performance metrics are available, the allocation is attempted in the

pool to which the device is bound.

If the pool initially chosen to allocate the data is full, FAST VP looks to

other pools contained within the FAST VP policy and allocates from

there. As long as there is space available in at least one of the pools

within the policy, all new extent allocations will be successful.

The allocation by policy feature is enabled at the Symmetrix array level

and applies to all allocations for all devices managed by FAST VP. The

Page 61: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

61

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

feature is either Enabled or Disabled (the default setting is Disabled).

When disabled, new allocations only come from the pool to which the

thin device is bound.

As a best practice, it is recommended that VP Allocation by FAST Policy

be enabled.

Note: For more information on the decision making process of the VP

allocation by FAST policy feature, refer to the Advance FAST VP features

section of the Implementing Fully Automated Storage Tiering for Virtual

Pools (FAST VP) for EMC Symmetrix VMAX Series Arrays Technical Note.

FAST VP tier configuration

A Symmetrix storage tier is a specification of a set of resources of the

same disk technology type (EFD, FC/SAS, SATA, or FTS storage),

combined with a given RAID protection type (RAID 1, RAID 5, RAID 6,

or Unprotected), and the same emulation (FBA or CKD).

Note: The Unprotected RAID type may only be applied to a tier residing on

an FTS-connected storage array.

FAST VP tiers can contain between one and four thin storage pools. To be

included in a FAST VP tier, a thin pool must contain data devices

configured on the same drive technology and emulation (and the same

rotational speed, in the case of FC, SAS, and SATA drives).

To combine multiple pools in a single tier, the pools must be of the same

RAID protection and configured on the same drive technology.

When a tier contains multiple pools, FAST VP movements into the tier

will be striped across all pools within the tier.

Drive-speed considerations

Two or more thin pools containing data devices configured on rotating

drives of different speeds, but of the same drive technology, may be

combined in a single VP tier. This is not recommended, however, as

there is the potential for uneven performance across the tier due to the

difference in drive speeds.

Drive-size considerations

Drive size is not a factor when adding data devices to a thin pool. For

Page 62: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

62

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

example, data devices configured on 300 GB FC 15 k drives can coexist in

a pool with data devices configured on 600 GB FC 15 k drives.

It is not recommended to combine drives of different sizes into the same

tier. The larger drives will ultimately contain more data than the smaller

drives. As such, they are likely to see a larger I/O workload than the

smaller drives, which may lead to different performance levels on data

devices across the tier.

If combining data devices on different drive sizes in the same storage tier

cannot be avoided, it is recommended to create separate pools for each

drive size, and then combine those pools into a single tier.

Note: For more information on best practices for configuring thin pools, refer

to the Best Practices for Fast, Simple Capacity Allocation with EMC

Symmetrix Virtual Provisioning Technical Note, available at

http://support.emc.com.

External tiers

External tiers can only be configured on storage configured for external

provisioning. Encapsulated storage is not supported by FAST VP.

When creating a tier on an external array, it is highly recommended that

the same best practices for the creation of internal thin pools and tiers be

followed for external pools and tiers. Primarily, external thin pools

should be configured on eDisks with similar performance profiles.

Note: For more information on best practices for configuring eDisks, refer to

the Design and Implementation Best Practices for EMC Symmetrix Federated

Tiered Storage (FTS) Technical Note, available at http://support.emc.com.

To that end, external thin pools should be created on eDisks from a

single external storage array. These eDisks should also be configured on

external LUNs of the same RAID protection and configured on the same

drive technology.

Page 63: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

63

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Similarly, if two or more external thin pools are added to an external tier,

the external pools should be configured on the same external array,

sharing the same RAID protection and drive technology type.

Greater care must be taken, however, in creating external pools and tiers.

The reason for this is that the Symmetrix array has no visibility into the

RAID protection or the drive technology of the external LUNs. As such,

there are no software protections to ensure that all the eDisks in a disk

group follow the rules that are applied to internal thin pools and tiers.

User-defined technology type

Depending on the configuration of the external tier (external RAID

protection and technology type), it may be possible that the tier can

provide a higher level of performance than an internal SATA, or possibly

an internal FC, tier.

In the case where it is expected that an external tier will perform at a

higher level than one or more internal tiers, the relative performance

expectation can be user-defined by specifying the technology type that

equates to the expected performance.

When the external tier is first added to a FAST VP policy, the Symmetrix

array executes a performance discovery in order to evaluate if the actual

performance is in line with the user-defined expectations.

Because of this performance discovery, it is recommended that external

tiers by added to a FAST VP policy at times when host workload is

relatively light. In this way, the full-discovery workload will compete

less with host workload.

Similarly, the external tier should also be added to the FAST VP policy as

early as is practical, prior to host applications accessing the external

storage.

FAST VP policy configuration

A FAST VP policy groups between one and four VP tiers and assigns an

upper-usage limit for each storage tier. The upper limit specifies the

maximum amount of capacity a storage group associated with the policy

can use on that particular tier. The upper-capacity usage limit for each

storage tier is specified as a percentage of the configured, logical capacity

Page 64: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

64

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

of the associated storage group.

The usage limit for each tier must be between 1 percent and 100 percent.

When combined, the upper-usage limit for all thin storage tiers in the

policy must total at least 100 percent, but may be greater than 100

percent.

Tier ranking

FAST VP ranks tiers within a policy based on known performance

models. Tiers are ranked based on the drive technology of the pools

within the each tier (EFD, FC, or SATA), as well as the location of the tier

(internal or external).

Tier usage limits

Creating a policy with a total upper-usage limit greater than 100 percent

allows flexibility with the configuration of a storage group, whereby data

may be moved between tiers without necessarily having to move a

corresponding amount of other data within the same storage group.

The default FAST VP policy would be to specify 100 percent for each of

the included tiers. Such a policy can provide the greatest amount of

flexibility to an associated storage group, as it allows 100 percent of the

storage group’s capacity to be promoted or demoted to any tier within

the policy, as appropriate.

Operationally, it may not be appropriate to deploy the 100/100/100

policy. There may be reasons to limit access to a particular tier within the

array.

As an example, it may be appropriate to limit the amount of a storage

group’s capacity that can be placed on EFD. This may be used to prevent

one single storage group, or application, from consuming all of the EFD

resources. In this case, a policy containing just a small percentage for

EFD would be recommended.

Page 65: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

65

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: Care must also be taken, however, to not set the policy percentage too

low for the higher-performing tiers in a policy. A percentage that is too low

may force FAST VP to leave active data on a lower tier, potentially

overloading that tier.

Similarly, it may be appropriate to restrict the amount of SATA capacity

a storage group may utilize. Some applications, which can become

inactive from time to time, may require a minimum level of performance

when they become active again. For such applications, a policy excluding

the SATA tier could be appropriate.

The best way to determine appropriate policies for a FAST VP

implementation is to examine the workload skew for the application

data to be managed by FAST VP. The workload skew defines an

asymmetry in data usage over time. This means a small percentage of the

data on the array may be servicing the majority of the workload on the

array. One tool that provides insight into this workload skew is Tier

Advisor.

Tier Advisor

Tier Advisor is a utility available to EMC technical staff that estimates

the performance and cost of mixing drives of different technology types

(EFD, FC, and SATA) within Symmetrix VMAX Family storage arrays.

Tier Advisor can examine performance data collected from Symmetrix,

VNX®, or CLARiiON® storage arrays and determine the workload skew

at the full LUN level. It can also estimate the workload skew at the sub-

LUN level.

With this information, Tier Advisor can model an optimal storage-array

configuration by enabling the ability to interactively experiment with

different storage tiers and storage policies, until achieving the desired

cost and performance preferences.

Thin device binding

Each thin device, whether under FAST VP control or not, may only be

bound to a single thin pool. By default, all host-write-generated

allocations, or user-requested pre-allocations, are performed from this

pool.

From an ease-of-management and reporting perspective, it is

Page 66: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

66

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

recommended that all thin devices be bound to a single pool within the

Symmetrix array.

Note: FAST VP data movements do not change the binding information for a

thin device.

In determining the appropriate pool, both performance requirements

and capacity management should be taken into consideration, as well as

the use of thin device pre-allocation and the system write-pending limit.

Note: Unless there is a very specific performance need, it is not

recommended to bind thin devices under FAST VP control to a thin pool in

the EFD tier.

Performance consideration

With VP allocation by FAST policy enabled, FAST VP attempts to

allocate new extents in the most appropriate tier. This is based on

available performance metrics for the thin device being allocated. Should

performance metrics not be available, the allocation comes from the pool

to which the thin device is bound.

As a result, the best practice recommendation is to bind all thin devices

to a pool within the FC tier in the policy.

Capacity-management consideration

Binding all thin devices to a single pool ultimately causes that single

pool to be oversubscribed. If the allocation by policy feature is not

enabled, this could potentially lead to issues as the pool fills up. Host

writes to unallocated areas of a thin device will fail if there is insufficient

space in the bound pool.

In an EFD, FC, SATA configuration, if the FC tier is significantly smaller

than the SATA tier, binding all thin devices to the SATA tier reduces the

likelihood of the bound pool filling up.

If performance needs require the thin devices to be bound to FC, the Pool

Reserved Capacity or the FAST VP policy configuration, or a

Page 67: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

67

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

combination of both, can be used to alleviate a potential pool-full

condition.

Note: Enabling VP allocation by FAST policy on the Symmetrix VMAX array

greatly alleviates the capacity-management consideration.

Pre-allocation

A way to avoid writes to a thin device failing due to a pool being fully

allocated is to pre-allocate the thin device when it is bound. However,

performance requirements of newly written data should be considered

prior to using pre-allocation.

When FAST VP performs data movements, only allocated extents are

moved. This applies not only to extents allocated as the result of a host

write, but also to extents that have been pre-allocated. These pre-

allocated extents will be moved even if no data has yet been written to

them.

Note: When moving pre-allocated, but unwritten, extents, no data is actually

moved. The pointer for the extent is simply redirected to the pool in the

target tier.

Pre-allocated, but unwritten, extents show as inactive and, as a result, are

demoted to the lowest tier included in the associated FAST VP policy.

When these extents are eventually written to, the write performance will

be that of the tier it has been demoted to.

A best-practice recommendation is to not pre-allocate thin devices

managed by FAST VP. Pre-allocation should only be used selectively for

those devices that can never tolerate a write failure due to a full pool.

System write-pending limit

FAST VP is designed to throttle back data movements, both promotions

and demotions, as the write-pending count approaches the system write-

pending limit. This throttling gives an even higher priority to host I/O to

ensure that tracks marked as write pending are destaged appropriately.

Note: By default, the system write-pending limit on a Symmetrix VMAX

array is set to 75 percent of the available cache.

Page 68: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

68

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

If the write-pending count reaches 60 percent of the write-pending limit,

FAST VP data movements stop. As the write-pending count decreases

below this level, data movements automatically restart.

A very busy workload running on SATA disks, with SATA disks at or

near 100 percent utilization, causing a high write-pending count, impacts

FAST VP from promoting active extents to FC or EFD tiers.

In an environment with a high write workload, it is recommended to

bind thin devices to a pool in the FC tier.

Note: If a large amount of write activity is being seen on the SATA tier, it

may be worth considering placing the DATA devices configured on SATA

into a separate dynamic cache partition. Doing so could prevent an increase

in the write-pending count and alleviate any impact on FAST VP and the

storage array as a whole.

Rebinding a thin device

It is possible to change the binding information for a thin device without

moving any of the current extent allocations for the device. This is done

by a process called rebinding.

Rebinding a thin device increases the subscription level of the pool to

which the device is bound, and decreases the subscription level of the

pool to which it was previously bound. However, the allocation levels in

both pools remain unchanged.

Note: When rebinding a device that is under FAST VP control, the thin pool

to which the device is being rebound must belong to one of the VP tiers

contained in the policy to which the device is associated.

Thin pool oversubscription

Thin pool oversubscription allows an organization to present larger than

needed devices to hosts and applications without having to purchase

enough physical drives to fully allocate all of the space represented by

the thin devices.

Page 69: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

69

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: For more information on oversubscription, refer to the Best Practices

for Fast, Simple Capacity Allocation with EMC Symmetrix Virtual

Provisioning Technical Note, available at http://support.emc.com.

In a FAST VP configuration, the capacity available for thin devices is

now spread across multiple pools. Even if there is no plan to

oversubscribe the available capacity in the Symmetrix array, there

typically is insufficient space in a single tier to accommodate the

configured thin device capacity.

By following the previous recommendation, binding all thin devices

being managed by FAST VP to a single thin pool inherently causes that

pool to be oversubscribed. As a result, the oversubscription limit for that

pool needs to be set to a value greater than 100 percent.

Available capacity

To determine the level to set the oversubscription limit, the total capacity

available to devices under the control of FAST VP should first be

calculated.

If the VP Allocation by FAST Policy feature is enabled, then 100 percent

of the capacity of each pool within the policy is available.

If the allocation by policy feature is not enabled, the following points

should be considered:

For pools with thin devices bound, 100 percent of the capacity is available.

For pools with no thin devices bound, 100 percent of the pool, less the capacity reserved by the PRC, is available.

Note: The pool reserved capacity value only affects FAST VP movements into

the pool. The PRC does not affect the ability for the thin devices to allocate

extents within the pool.

Oversubscription limits

After determining the capacity available to FAST VP, an

oversubscription limit can then be calculated for the pool devices that are

going to be bound to it.

To ensure the configured capacity of the array is not oversubscribed, the

limit can be calculated by dividing the capacity of the thin pool being

Page 70: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

70

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

used for binding into the available capacity of all the pools. This value is

then multiplied by 100 to get a percent value.

For example, consider a configuration with a 1 TB EFD pool, a 5 TB FC

pool, and a 10 TB SATA pool. The total available capacity is 16 TB. If all

thin devices are bound to FC, the oversubscription limit could be set to

320% — (16/5)*100.

This value can be set higher if the intention is to initially oversubscribe

the configured physical capacity of the array, and then add storage on an

as-needed basis.

For thin pools where devices will never be bound, the subscription limit

can be set to 0 percent. This prevents any accidental binding or migration

to that pool.

RAID protection considerations

When designing a Virtual Provisioning configuration, and particularly

when choosing a corresponding RAID protection strategy, both device-

level performance and availability implications should be carefully

considered.

Note: For more information on these considerations, refer to the Best

Practices for Fast, Simple Capacity Allocation with EMC Symmetrix Virtual

Provisioning Technical Note, available at http://support.emc.com.

FAST VP does not change these considerations and recommendations

from a performance perspective. What FAST VP does change, however,

is that a single thin device can now have its data spread across multiple

tiers of varying RAID protection and drive technology within the array.

Because of this, the availability of an individual thin device is not based

just on the availability characteristics of the thin pool to which the device

is bound. Instead, availability is based on the characteristics of the tier

with the lowest availability.

Page 71: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

71

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

While performance and availability requirements ultimately determine

the configuration of each tier within the Symmetrix array, it is

recommended as a best practice to choose RAID 1 or RAID 5 (preferably

3+1) protection on EFDs. The faster rebuild times of EFDs provide higher

availability for these protection schemes on that tier.

Also, it is recommended to use either RAID 1 or RAID 6 (preferably 6+2)

on the SATA tier. This is due to the slower rebuild times of the SATA

drives (compared to EFD and FC), and the increased chance of a dual-

drive failure, leading to data unavailability with RAID 5 protection.

For the FC tier, RAID 1 is the recommended protection level. Mirrored

data devices on FC pools provide a higher level of performance than

both RAID 5 and RAID 6, particularly for write workloads. Availability

of RAID 1, in regard to a dual-drive failure, is also greater than RAID 5.

To obtain the best availability numbers for RAID 1 on FC, the use of

lower-capacity drives is recommended.

Note: For new environments being configured in anticipation of

implementing FAST VP, or for existing environments adding more tiers, it is

highly recommended that an EMC representative be engaged to assist in

determining the appropriate RAID protection schemes for each tier.

Drive configuration

For most customer configurations, the best-practice configuration

guidelines recommend an even, balanced distribution of physical disks

across the disk adapters (DAs) and DA CPUs.

This is of particular relevance for EFDs, which are each able to support

thousands of I/Os per second, and, therefore, are able to create a load on

the DA equivalent to that of multiple hard disk drives.

An even distribution of I/O across all DA CPUs is optimal to maximize

their capability and the overall capability of the Symmetrix array.

However, there are scenarios where it is not appropriate to evenly

distribute disk resources. An example of this is when the customer

desires partitioning and isolation of disk resources to separate customer

environments and workloads within the VMAX array.

Generally, it is appropriate to configure more, smaller drives, than fewer,

larger drives for both EFD and FC tiers. Doing so spreads I/O load as

Page 72: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

72

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

wide as possible across the Symmetrix back-end.

Storage group priority

When a storage group is associated with a FAST policy, a priority value

must be assigned to the storage group. This priority value can be

between 1 and 3, with 1 being the highest priority. The default is 2.

The best-practice recommendation is to use the default priority of 2 for

all storage groups associated with FAST VP policies. These values may

then be modified if it is seen, for example, that a high-priority

application is not getting sufficient resources from a higher tier.

SRDF

FAST VP has no restrictions in its ability to manage SRDF devices,

however, what must be considered is that data movements are restricted

to the array upon which FAST VP is operating. By default, there is no

coordination of data movements; FAST VP acts independently on both

the local and remote arrays. However, enabling FAST VP SRDF

coordination allows R1 performance metrics to be used when making

promotion-and-demotion decisions for data belonging to an R2 device.

Note: For more information on the operation of FAST VP SRDF coordination,

refer to Implementing Fully Automated Storage Tiering for Virtual Pools (FAST

VP) for EMC Symmetrix VMAX Series Arrays Technical Note, available at

http://support.emc.com.

As a general best practice, FAST VP should be employed for both R1 and

R2 devices, particularly if the remote R2 array has also been configured

with tiered storage capacity. Where possible, similar FAST VP tiers and

policies should be configured at each site.

When available, FAST VP SRDF coordination should be enabled for each

storage group containing SRDF devices that is associated with a FAST

VP policy.

Page 73: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

73

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Each SRDF configuration, however, presents its own unique behaviors

and workloads. Prior to implementing FAST VP in an SRDF

environment, the information in the following sections should be

considered.

Note: The following sections assume that SRDF is implemented with all

Symmetrix arrays configured for Virtual Provisioning (all SRDF devices are

thin devices) and capable of running FAST VP.

FAST VP behavior

For SRDF R1 devices, FAST VP promotes and demotes extent groups

based on the read-and-write activity experienced on the R1 devices.

Meanwhile, by default, the SRDF R2 devices typically only experience

write activity during normal operations. As a result, FAST VP is likely to

promote only R2 device extents that are experiencing writes.

If there are R1 device extents that only experience read activity, no

writes, then the corresponding extents on the R2 devices will see no I/O

activity. This will likely lead to these R2 device extents being demoted to

the SATA tier, assuming this was included in the FAST VP policy.

SRDF coordination changes this default behavior by periodically

transmitting FAST VP performance metrics collected for R1 devices

across the SRDF link to the corresponding R2 devices. On the R2 device,

the R1 performance metrics are merged with the actual R2 metrics.

Movement decisions made by FAST VP for these R2 devices are then

based on the merged performance metrics. In this way, read activity on

the R1 device can be reflected on the R2.

SRDF operating mode

EMC best practices, for both synchronous and asynchronous modes of

SRDF operation, recommend implementing a balanced configuration on

both the local (R1) and remote (R2) Symmetrix arrays. Ideally, data on

each array would be located on devices configured with the same RAID

protection type, on the same drive type.

As FAST VP operates independently on each array, and also promotes

and demotes data at the sub-LUN level, there is no guarantee that such a

balance may be maintained.

Page 74: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

74

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

In SRDF synchronous (SRDF/S) mode, host writes are transferred

synchronously from R1 to R2. These writes are only acknowledged by

the host when the data has been received into cache on the remote R2

array. These writes to cache are then destaged asynchronously to disk on

the R2 array.

Note: For more information on SRDF/S, see the EMC Solutions Enabler

Symmetrix SRDF Family CLI Product Guide, available at

http://support.emc.com.

In an unbalanced configuration, where the R2 data resides on a lower-

performing tier than on the R1, performance impact may be seen at the

host if the number of write pendings builds up and writes to cache are

delayed on the R2 array.

With FAST VP, this typically does not cause a problem, as the

promotions that occur on the R2 side are the result of write activity.

Areas of the thin devices under heavy write workload are likely to be

promoted and maintained on the higher-performing tiers on the R2

array.

To avoid potential problems with the configuration becoming

unbalanced, it is recommended that FAST VP SRDF coordination be

enabled in SRDF/S environments.

In SRDF asynchronous (SRDF/A) mode, host writes are transferred

asynchronously in predefined time periods or delta sets. At any given

time, there are three delta sets in effect: The capture set, the

transmit/receive set, and the apply set.

Note: For more information on SRDF/A, see the EMC Solutions Enabler

Symmetrix SRDF Family CLI Product Guide, available at

http://support.emc.com.

A balanced SRDF configuration is more important for SRDF/A, as data

cannot transition from one delta set to the next until the apply set has

Page 75: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

75

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

completed destaging to disk. If the data resides on a lower-performing

tier on the R2 array, compared to the R1, then the SRDF/A cycle time

may elongate and eventually cause the SRDF/A session to drop.

This is similar to SRDF/S mode in most environments, so this may not

be a large issue, as the data under write workload is promoted and

maintained on the higher-performing tiers.

To avoid potential problems with the configuration becoming

unbalanced, it is strongly recommended that FAST VP SRDF

coordination be enabled in SRDF/A environments.

SRDF/A DSE (delta set extension) should be considered to prevent

SRDF/A sessions from dropping in situations where writes propagated

to the R2 array are being destaged to a lower tier, potentially causing an

elongation of the SRDF/A cycle time.

Note: For more information on SRDF/DSE, see the Best Practices for EMC

SRDF/A Delta Set Extension Technical Note, available at

http://support.emc.com.

SRDF failover

As FAST VP works independently on both the R1 and R2 arrays, it

should be expected that the data layout will be different on each side if

SRDF coordination is not enabled. When an SRDF failover operation is

performed, and host applications are brought up on the R2 devices, it

should then also be expected that the performance characteristics on the

R2 will be different from those on the R1.

In this situation, it takes FAST VP some period of time to adjust to the

change in workload and start promotion-and-demotion activities based

on the mixed read-and-write workload.

To better prepare for a failover, and to shorten the time period to achieve

an adjustment to the new workload, FAST VP SRDF coordination should

be enabled.

Note: FAST VP performance metrics are only transmitted from R1 to R2.

When failed over, and with applications are running on R2 devices,

performance metrics will not be sent from R2 to R1.

Page 76: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

76

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Personality swap

An SRDF personality swap follows a similar pattern of behavior to an

SRDF failover scenario. Applications are brought up on the devices that

were formerly R2 devices.

For performance metrics to be transmitted from the new R1 devices to

the corresponding R2 devices, SRDF coordination needs to be enabled on

the storage groups containing the R1 devices.

SRDF coordination can be enabled in advance on a storage group

containing R2 devices. However, the setting only takes effect when a

personality swap is performed, converting the R2 devices to R1.

SRDF bi-directional

Best-practice recommendations change slightly in a bi-directional SRDF

environment. In this case, each Symmetrix array has both R1 and R2

devices configured.

With SRDF coordination enabled, it is possible that data belonging to an

R2 device could displace data belonging to an R1 device on a higher tier.

This may happen if the R2 device’s corresponding R1 device has a higher

workload than the local R1 device.

In this scenario, it is recommended to reserve the EFD tier for R1 device

usage. This can be done by excluding any EFD tiers from the policies

associated with the R2 devices.

Should a failover be performed, then the EFD tier can be added

dynamically to the policy associated with the R2 devices. Alternatively,

the storage group containing the R2 devices could be reassociated with a

policy containing EFD.

Note: This recommendation assumes that the lower tiers are of sufficient

capacity and I/O capability to handle the expected SRDF write workload on

the R2 devices.

Page 77: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

77

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

EFD considerations

If there is a smaller EFD configuration on the remote array, it is

recommended not to include the EFD tier in the FAST VP policies on the

R2 array. Similarly, if the EFD configuration on the R2 array does not

follow the best-practice guideline of being balanced across DAs, do not

include the EFD tier on the R2 side.

Note: This recommendation assumes that the lower tiers are of sufficient

capacity and I/O capability to handle the expected SRDF write workload on

the R2 devices.

FAST VP Compression

FAST VP Compression automates VP compression for thin devices that

are managed by FAST VP.

The behavior of FAST VP Compression is controlled by two

configuration parameters: FAST VP Time to Compress and FAST VP

Compression Rate. FAST VP Compression also requires that at least one

tier within the policy contains at least one thin pool that has been

enabled for VP compression.

FAST VP Time to Compress

The FAST VP Time to Compress parameter enables the use of FAST VP

Compression and sets the inactivity threshold for when data becomes

eligible for compression.

The time to compress can be configured to be between 40 and 400 days,

as well as never. The default is never, meaning FAST VP will not

automatically compress any data.

Prior to enabling FAST VP Compression, users should have an

understanding of their typical business cycle, and the frequency at which

all data may be considered active. For example, end-of-month reporting

or batch processing.

As a best practice, the time to compress should initially be set to a

minimum of 40 days. This ensures data needed for any potential end-of-

month activity during the month the data was created in will not be

compressed prior to the end of that month.

If the typical access cycle for data stretches over a business quarter (three

Page 78: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

78

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

months), it would then be appropriate to set the time to compress to a

minimum of 100 days.

FAST VP Compression Rate

The FAST Compression Rate is a quality of service setting for the

automated VP compression performed by FAST VP. It affects the

aggressiveness with which data is compressed, after it has been inactive

beyond the time-to-compress value.

The compression rate can be configured to be between 1 and 10, with 1

being the most aggressive.

Setting the compression rate to 1 allows FAST VP to attempt to compress

data as quickly as possible. However, if there is a large amount of data

to be compressed, this may add additional overhead to the back end,

potentially impacting the host I/O response times.

Setting the compression rate to the least aggressive value, 10, causes

FAST VP to greatly reduce the amount of data that will be compressed at

any given time. This setting is less likely to impact the host response

time, but it will take longer for the compression to complete.

The default value for the compression rate is 5. However, the best-

practice recommendation for the initial deployment of FAST VP is to

start with a more conservative value, perhaps 7 or 8.

At a later date, when the level of impact compression has on the array

has been observed, the compression rate can potentially be increased to a

more aggressive setting.

Compression-enabled tiers

In order for data to be automatically compressed it must be allocated in a

thin pool that has been enabled for compression. Compression can be

enabled on any pool, or any location (internal or external), and any

technology type (EFD, FC, SATA).

Typically, any data that has been inactive for a period of time longer

than the time to compress will have already been demoted to the lowest

Page 79: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

79

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

tier in the policy. As such, the recommendation for FAST VP

Compression is that only the lowest tier configured for the Symmetrix

VMAX array, internal or external, should be enabled for compression.

Migration to a FAST VP environment

Migrating from a non-tiered environment to a tiered environment can

provide several challenges. Considerations must be made regarding

both capacity and performance.

From a capacity perspective, not all the data on the single-tiered storage

array may fit into any one tier on the multi-tiered array, so migrations

must be planned very carefully.

When performing a migration to thin devices, it is possible that the thin

devices may become fully allocated as a result of the migration. Because

of this, the target thin devices being migrated to should be configured in

such a way that 100 percent of the space required to complete the

migration is available.

Note: The possibility of a migration fully allocating the target thin devices

can be mitigated by employing existing zero-space reclamation tools. For

more information refer to “Zero space reclamation” on page 84.

From a performance perspective, there are two different aspects to be

considered. The first is the performance of the data on the target array

when the migration has completed. The second aspect is the

performance of the migration itself.

Any migration into a multi-tiered environment should consider whether

the target tier can provide a sufficient level of performance to the

applications migrated until such time that FAST VP can appropriately

redistribute the data across multiple tiers. Similarly, the target tiers

should be able to absorb a large amount of writes at a performance level

that allows the migration to complete in an acceptable timeframe.

Migration planning

When migrating into a multi-tiered FAST VP environment, there are

three primary migration methods available, each of which has options to

ensure the necessary space is available to the target thin devices. The

three methods are:

Page 80: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

80

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Migration to a single tier

Migration to multiple tiers

Migration to a FAST VP policy

The following sections discuss the options available to ensure sufficient

space is available to complete the migration successfully for each

migration method.

Migration to a single tier

When migrating to a thin device, the device can only be bound to a

single pool. Typically, all thin devices being used for a single application

are bound to the same pool. In this scenario, the target pool needs to

have sufficient space to allow each of the thin devices to be fully

allocated in order to guarantee the migration will be completed

successfully.

If the target pool has less free space than is needed, the pool reserve

capacity (PRC) can be increased prior to the migration in order to make

the space available. Increasing the PRC makes less space available in the

pool for use by FAST VP. As a result, if the PRC of a pool is increased to

a level greater than the percent of free space in the pool, FAST VP

actively starts moving data out of the pool in order to bring the percent

of free space in line with the new PRC value.

For example, if a pool is 200 TB in size and has a PRC of 10 percent

defined, 180 TB is available for use by FAST VP. Increasing the PRC to 20

percent means that only 160 TB is available to FAST VP. If the pool was

170 TB allocated to devices managed by FAST VP prior to the increase in

PRC, FAST VP would attempt to relocate 10 TB worth of data to decrease

the allocation level to 160 TB.

Prior to starting a migration, the PRC should be changed to a level that

reserves the required capacity in the pool the target devices are bound to.

When migrating to a single tier, the target devices should not be

associated with a FAST VP policy until after the migration has

completed.

Page 81: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

81

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Note: Depending on the amount of data to be moved, and the existing level

of activity on the array, it may take a significant amount of time to move data

from the pool to reach the new PRC level. As such, any decision to increase

the PRC for the purposes of a migration should be made well in advance of

the migration being executed.

Migration to multiple tiers

If there is insufficient space in a single pool to accept all the data to be

migrated, it may be necessary to involve multiple thin pools on the

target array.

Migration to multiple tiers implies that the target thin devices are not all

bound to a single pool. Rather, multiple target pools are used to

complete the migration.

Individual target devices should be bound to the most appropriate pool

to complete the migration of each device.

The appropriateness of the pool used can be based on the amount of free

capacity of the pool, but also on the performance requirements of the

devices being migrated. For example, the most active devices can be

migrated to a pool, or pools, in the FC tier, while the remaining, less

active devices are bound to a pool, or pools, in the SATA tier.

Ideally, performance data for the devices being migrated is available

prior to the planning of the migration. If this information is not available,

a determination should be made as to which are the most performance-

sensitive devices being migrated.

Care should be exercised when migrating a large amount of data to the

SATA tier. If the tier is configured with RAID 6 protection, the migration

activity will generate a higher level of write activity to the disk drives.

This could impact both the performance of the migration as well as the

overall performance of the target storage array.

When migrating to multiple tiers, the target devices should not be

associated with a FAST VP policy until the migration has completed.

Migration to a FAST VP policy

If there is insufficient space in a single target pool to accept all the

migration data, the VP Allocation by FAST VP Policy feature may be

used to aid the migration. Using allocation by policy, the target thin

Page 82: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

82

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

devices can all be bound to a single pool. Prior to starting the migration,

the target devices should also be associated with a FAST VP policy.

With the allocation by policy feature enabled, if the pool to which the

target devices are bound fills up during the migration, allocations spills

over to other pools in the FAST VP policy. As long as there is sufficient

space in the pools within the FAST VP policy, all the data will be

migrated successfully.

As with a migration to multiple tiers, care should be taken when

migrating a large amount of data to drives configured with RAID 6

protection.

FAST VP also cannot take the performance requirements of the data

being migrated into account, since no performance history is available.

This means that highly active data may end up being placed on a lower-

performing tier while inactive data may be placed on a higher-

performing tier. For some period of time after the migration, the

migrated data may suffer from degraded performance until such time

that FAST VP has the opportunity to optimize the layout of the data for

the actual workload.

When using the migration to a FAST VP policy strategy, the target

devices should be associated with a policy prior to the migration

starting.

Zero space reclamation

Because of the potential for a migration to fully allocate the target thin

device, Virtual Provisioning zero space reclamation should be used

following the migration to deallocate zero data copied during the

migration. It is recommended to pin thin devices managed by FAST VP

prior to starting a zero-space-reclamation task on them.

If either SRDF or Open Replicator is being used, the zero-detection

capability of each should be used during the migration. This can prevent

target thin devices from being fully allocated during the migration itself.

Page 83: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

83

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Upgrade planning

When planning a drive upgrade in a multi-tier, FAST VP environment,

the biggest question is to which tier capacity should be added. In trying

to determine that, an understanding is required of the increase in overall

workload for the storage array that will come as a result of the added

capacity.

If it is anticipated that not much additional workload will be generated,

and the existing configuration is providing satisfactory performance, it

may be appropriate to simply add more capacity to the SATA tier.

Note: Adding capacity to only one tier will change the relative percentage of

data that can live on the other tiers, which may cause some of the current

workloads to be migrated to new (perhaps lower) tiers in the future.

If a large increase in workload is anticipated, however, the overall

performance of the array and the performance of each storage tier need

to be considered before deciding where capacity should be added.

The following sections describe some key performance indicators (KPIs)

available in Performance Analyzer that can be viewed, along with some

general considerations, to help determine how a drive upgrade should

be planned.

System

From the system level, the overall activity level of the array should be

examined, including the total number of I/Os being performed, as well

as the read/write workload mix. Performance metrics should also be

examined to differentiate between the front-end workload and the back-

end workload.

Performance metrics that should be looked at in Performance Analyzer

include:

Host IOs/sec

% Reads

% Writes

Host MBs/sec

BE IOs/sec

BE Reads/sec

Page 84: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

84

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

BE Writes/sec

% Cache write pending

Systems with a high number of writes, or high percentage of writes,

should consider adding drives to the higher-performing tiers, EFD and

FC. If FC is configured as RAID 1, this may be the most appropriate tier

to expand.

Note: Systems with a high write-pending level are likely to need additional

cache installed as a part of supporting any significant increase in the usable

capacity of the array.

Virtual Pool tiers

The workload and behavior of each individual tier should also be

examined, primarily at the back end. Metrics that should be looked at in

Performance Analyzer include:

BE Reqs/sec

BE % Reads

BE % Writes

BE Avg Response Time

Total Pool Capacity (GB)

Used Pool Capacity (GB)

Note: If a tier contains multiple pools, these metrics should be examined for

each individual pool in the tier.

Tiers with higher response times, or a higher used-capacity percentage,

are most likely to benefit from being expanded. Expanding a higher tier

may reduce the workload on a lower tier as more active data is promoted

upwards.

Back-end directors

One of the best-practice recommendations for FAST VP is to have a

balanced configuration on the back end of the VMAX storage array. As a

result, the workload on each back-end director (DA) should be examined

Page 85: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

85

Best practice considerations and recommendations

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

to determine if a similar workload is being seen on each DA. If not, the

upgrade plan should include drives to achieve a better balance.

The relevant performance metrics for the DAs include:

IOs/sec

% Reads

% Writes

% Busy

Ideally, each logical DA should have a similar workload profile for each

of these metrics. Should any imbalance exist, the drive upgrade should

include the correct mix of drives to achieve balance on the DAs.

Page 86: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

86

Conclusion

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Conclusion

EMC Symmetrix VMAX FAST VP for Virtual Provisioning environments

automates the identification of active or inactive application data for the

purposes of reallocating that data across different performance/capacity

tiers within an array. FAST VP proactively monitors workloads at both

the LUN and sub-LUN level in order to identify busy data that would

benefit from being moved to higher-performing drives. FAST VP also

identifies less-busy data that could be moved to higher-capacity drives,

without existing performance being affected. This promotion/demotion

activity is based on policies that associate a storage group to multiple

drive technologies, or RAID protection schemes, by way of thin storage

pools, as well as the performance requirements of the application

contained within the storage group. Data movement executed during

this activity is performed non-disruptively, without affecting business

continuity and data availability.

Page 87: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

87

Appendix A: FAST VP state

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Appendix A: FAST VP state

There are five possible states that the FAST controller can be reported in.

These are:

Enabled: All FAST VP functions are performed. Performance data collection, performance data analysis, data-movement request generation, and data-movement execution.

Disabled: Only performance data collection is performed. Data analysis is not performed, and data movement is not executed.

Disabling: The transition of the FAST controller from Enabled to Disabled.

DisabledwithError: The FAST controller has stopped operation due to an internal error. Statistics collection and FAST VP performance data movements continue to be performed, however, FAST VP compliance movements are not performed.

Degraded: FAST VP can perform some or all of its functions. However, it cannot perform each function fully.

Enabled state

When the state of the FAST controller is queried, and the state is

Enabled, the current activity being performed by the controller is also

displayed. Valid activities include:

Idle: The FAST controller is currently idle.

RunningPlan: There are currently active data-movement tasks running, moving thin device data between tiers.

Degraded state

When the state of the FAST controller is Degraded, a reason code is

displayed when the FAST state is queried, and it indicates the cause of

the degraded state.

These reason codes include:

Invalid Swap/Performance time windows: At least one of the defined time windows is invalid. To correct, each time window should be checked, and any invalid time windows should be deleted or modified.

Invalid device attributes: One or more storage groups have an invalid priority in a FAST policy. To correct, each storage group’s priority should be checked in the FAST policy they are associated

Page 88: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

88

Appendix A: FAST VP state

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

with. Any invalid priority should be modified to a valid value.

Invalid FAST parameters: One or more of the FAST controller configuration settings are invalid. To correct, each configuration setting should be checked and set to a valid value.

Performance time window is not present or does not extend into the future: No performance time window, default or user-defined, exists, or any that do exist have expired. To correct, a valid, inclusion performance time window should be created.

FAST thin move time window is not present or does not extend into the future: No thin data movement time window, default or user-defined, exists, or any that do exist have expired. To correct, a valid, inclusion thin data movement time window should be created.

FAST VP compliance movement failed: The most recent attempt to perform a FAST VP compliance movement was not successful. EMC customer service should be contacted to investigate the reason for the failure. If a subsequent attempt to perform a compliance movement is successful, the degraded state is cleared.

FAST VP performance movement policy update failed: The most recent attempt to generate a data-movement policy failed. EMC customer service should be contacted to investigate. If a subsequent attempt to generate a movement policy is successful, the degraded state is cleared.

FAST VP is not licensed: An entitlement file including FAST VP has not been loaded to the Symmetrix array. To correct, the appropriate entitlement file should be obtained from EMC and loaded to the Symmetrix array.

Statistics collection is failing for thin devices - No Performance movement will happen: Performance statistics are not being collected for thin devices under FAST VP control. EMC Customer Service should be contacted to investigate. If a subsequent attempt to collect statistics is successful, the degraded state is cleared.

Timed out attempting to communicate with the FAST controller: Either the FAST controller running on the service processor is unavailable, or the service processor itself is unavailable. EMC Customer Service should be contacted to investigate.

Page 89: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

89

Appendix B: Feature support

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Appendix B: Feature support

The following table describes the minimum Enginuity and management

interface levels needed to support various FAST VP features.

Feature Enginuity Management interface

FAST VP (Base) 5875.135.91 Solutions Enabler V7.3

SMC 7.3

Unisphere for VMAX 1.0

Setting PRC per pool 5875.198.38 Solutions Enabler V7.3.1

SMC 7.3.1

Unisphere for VMAX 1.0

VP Allocation by FAST

Policy

5876.82.57 Solutions Enabler V7.4

Unisphere for VMAX 1.0

FAST VP SRDF

coordination

5876.82.57 Solutions Enabler V7.4

Unisphere for VMAX 1.0

FAST VP SRDF

coordination for multi-site

SRDF

5876.229.145 Solutions Enabler V7.6

Unisphere for VMAX 1.6

External tier (FTS) 5876.82.57 Solutions Enabler V7.4

Unisphere for VMAX 1.0

Storage group reassociation 5876.82.57 Solutions Enabler V7.4

Unisphere for VMAX 1.0

FAST VP for CKD 5876.82.57 Solutions Enabler V 7.4

Unisphere for VMAX 1.0

FAST VP for IBM i 5876.82.57 Solutions Enabler V 7.4

Unisphere for VMAX 1.0

Page 90: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

90

Appendix B: Feature support

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

FAST VP Compression 5876.159.102 Solutions Enabler V7.5

Unisphere for VMAX 1.5

Four tiers in a FAST VP

policy

5876.159.102 Solutions Enabler V 7.5

Unisphere for VMAX 1.5

User-defined FTS tier 5876.159.102 Solutions Enabler V7.5

Unisphere for VMAX 1.5

Page 91: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

91

Appendix C: Best practices quick reference

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

Appendix C: Best practices quick reference

The following provides a quick reference to the general best-practice recommendations

for planning the implementation of a FAST VP environment.

The best practices documented are based on features available in Enginuity 5876.229.145,

Solutions Enabler V7.6, and Unisphere for VMAX 1.6.

Note: For more details on these recommendations, and other considerations,

see “Best practice considerations” on page 56.

FAST VP configuration parameters

FAST VP includes multiple configuration parameters that control its behavior. This

appendix provides a brief summary of each of the best-practice recommendations. For

more details on each recommendation, click on a heading to redirect to the appropriate

section of the document.

Performance time window

Use the default performance time window to collect performance metrics 24 hours a day,

every day.

Data movement time window

Create a data movement window to allow data movement for the same period of time

that the performance time windows allow data collection.

Workload Analysis Period (WAP)

The best-practice recommendation for the Workload Analysis Period is to match it to the

performance time window to cover one calendar week.

Initial Analysis Period (IAP)

At the initial deployment of FAST VP, set the Initial Analysis Period to 168 hours to

ensure that a typical weekly workload cycle is seen. After this period has passed, the

value can be dropped to 24 hours.

FAST VP Relocation Rate (FRR)

For the initial deployment of FAST VP, start with a conservative value for the relocation

rate, perhaps 7 or 8. At a later date, the FRR can be gradually lowered to a more

aggressive level.

Pool Reserved Capacity (PRC)

For individual pools with bound thin devices, set the PRC based on the lowest allocation

Page 92: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

92

Appendix C: Best practices quick reference

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

warning level for that thin pool. For pools with no bound thin devices, set the PRC to 1

percent.

VP Allocation by FAST Policy

VP Allocation by FAST Policy should be enabled.

FAST VP Time to Compress

The FAST VP Time to Compress should initially be set to a minimum of 40 days to

account for a typical monthly workload cycle, or 100 days to account for a three-month

cycle.

FAST VP Compression Rate

For the initial deployment of FAST VP, start with a more conservative value for the

compression rate, perhaps 7 or 8. At a later date, the compression rate can potentially be

increased to a more aggressive setting.

FAST VP Compression

For FAST VP Compression, it is recommended that only the lowest tier configured for

the Symmetrix VMAX array, internal or external, be enabled for compression.

FAST VP policy configuration

The ideal FAST VP policy would be 100 percent EFD, 100 percent FC, and 100 percent

SATA. While ideal, operationally it may not be appropriate to deploy the 100/100/100

policy. There may be reasons to limit access to a particular tier within the array.

The best way to determine the appropriate policies for a FAST VP implementation is to examine the workload skew for the application data to be managed by FAST VP. This can be done by using a tool such as Tier Advisor.

Thin device binding

The best-practice recommendation is to bind all thin devices to a pool within the FC tier.

VP allocation by FAST policy should be enabled.

It is not recommended to bind thin devices, under FAST VP control, to the EFD tier.

A best-practice recommendation is to not pre-allocate thin devices managed by FAST VP.

RAID protection considerations

For EFD, choose RAID 5 (3+1) protection.

Page 93: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

93

Appendix C: Best practices quick reference

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

For FC, choose RAID 1 protection.

For SATA, choose RAID 6 (6+2) protection.

Drive configuration

Where possible, balance physical drives evenly across DAs. Configure more, smaller

EFDs, rather than fewer, larger EFDs, to spread the I/O load as wide as possible.

Storage group priority

The best-practice recommendation is to use the default priority of 2 for all storage groups

associated with FAST VP policies.

SRDF

As a general best practice, FAST VP should be employed for both R1 and R2 devices,

particularly if the remote R2 array has also been configured with tiered storage capacity.

Similar FAST VP tiers and policies should be configured at each site.

FAST VP SRDF coordination should be enabled for storage groups containing SRDF R1

devices.

Page 94: Docu31003 FAST VP for Symmetrix VMAX Theory and Best Practices for Planning and Performance

94

References

Implementing FAST VP for EMC Symmetrix VMAX Series Arrays Technical Notes

References

EMC Solutions Enabler Symmetrix Array Controls CLI Product Guide

EMC Solutions Enabler Symmetrix Array Management CLI Product Guide

EMC Solutions Enabler Symmetrix CLI Command Reference HTML Help

EMC Solutions Enabler Installation Guide

EMC Symmetrix VMAX Series Product Guide

Implementing Fully Automated Storage Tiering for Virtual Pools (FAST

VP) for EMC VMAX Series Arrays

Best Practices for Fast, Simple Capacity Allocation with EMC Symmetrix Virtual Provisioning Technical Note

z/OS and Virtual Provisioning Best Practices

Design and Implementation Best Practices for EMC Symmetrix Federated Tiered Storage (FTS) Technical Note

Best Practices for Nondisruptive Tiering via EMC Symmetrix Virtual LUN Technical Note

Copyright © 2013 EMC Corporation. All Rights Reserved.

EMC believes the information in this publication is accurate as of its publication date. The

information is subject to change without notice.

THE INFORMATION IN THIS PUBLICATION IS PROVIDED "AS IS." EMC

CORPORATION MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND

WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION, AND SPECIFICALLY

DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A

PARTICULAR PURPOSE.

Use, copying, and distribution of any EMC software described in this publication requires

an applicable software license.

For the most up-to-date listing of EMC product names, see EMC Corporation Trademarks

on EMC.com.

All other trademarks used herein are the property of their respective owners.