Lab Validation Report - c368768.ssl.cf1.rackcdn.com Validation Report Springpath Data Platform Enterprise Data Management with Cloud Economics and Maximum Simplicity By Tony Palmer,

Lab Validation Report Springpath Data Platform

Enterprise Data Management with Cloud Economics and Maximum Simplicity

By Tony Palmer, Senior Lab Analyst

February 2015 © 2015 by The Enterprise Strategy Group, Inc. All Rights Reserved.

Lab Validation: Springpath Data Platform 2

© 2015 by The Enterprise Strategy Group, Inc. All Rights Reserved.

Contents

Introduction .................................................................................................................................................. 3 Challenges ................................................................................................................................................................. 3 Springpath Data Platform for VMware ..................................................................................................................... 4

ESG Lab Validation ........................................................................................................................................ 6 Getting Started ......................................................................................................................................................... 6 High Availability ...................................................................................................................................................... 12 Mixed Workload Performance ............................................................................................................................... 14

ESG Lab Validation Highlights ..................................................................................................................... 19

Issues to Consider ....................................................................................................................................... 19

The Bigger Truth ......................................................................................................................................... 20

Appendix ..................................................................................................................................................... 21

All trademark names are property of their respective companies. Information contained in this publication has been obtained by sources The Enterprise Strategy Group (ESG) considers to be reliable but is not warranted by ESG. This publication may contain opinions of ESG, which are subject to change from time to time. This publication is copyrighted by The Enterprise Strategy Group, Inc. Any reproduction or redistribution of this publication, in whole or in part, whether in hard-copy format, electronically, or otherwise to persons not authorized to receive it, without the express consent of The Enterprise Strategy Group, Inc., is in violation of U.S. copyright law and will be subject to an action for civil damages and, if applicable, criminal prosecution. Should you have any questions, please contact ESG Client Relations at 508.482.0188.

ESG Lab Reports

The goal of ESG Lab reports is to educate IT professionals about data center technology products for companies of all types and sizes. ESG Lab reports are not meant to replace the evaluation process that should be conducted before making purchasing decisions, but rather to provide insight into these emerging technologies. Our objective is to go over some of the more valuable feature/functions of products, show how they can be used to solve real customer problems and identify any areas needing improvement. ESG Lab's expert third-party perspective is based on our own hands-on testing as well as on interviews with customers who use these products in production environments. This ESG Lab report was sponsored by Springpath.



Introduction

This ESG Lab Validation documents the results of hands-on evaluation and testing of Springpath’s Data Platform software. This report focuses on the simplicity of deployment and management, resource efficiency, and scalability of the Springpath software. ESG Lab focused on the ability of Springpath to extend beyond the capabilities of traditional converged systems; eliminate silos of compute, networking, and storage resources; and significantly reduce total cost of ownership.

Challenges

Server virtualization is driving widespread and fundamental change in business. IT organizations struggle to meet the exponentially increasing demand for network and storage resources in support of virtualized deployments. Today’s traditional IT environment is complex, costly, and inflexible, inhibiting IT staff from effectively supporting the business. The challenges this presents to IT include:

Islands of Functionality – IT consists of many islands of functionality and resources, which leads to low utilization and high labor costs. In the early 90s, primary storage went through a horizontal consolidation wave that centralized the islands of primary storage devices that had been locked inside physical servers into a shared device. VMware has done the same for servers. The challenge is bringing a wave of vertical consolidation to the data center; basically consolidating servers, storage, and multiple, special-purpose appliances and software applications into a single, virtualized, and highly utilized shared resource pool. In summary, the challenge is to assimilate primary storage, servers, backup appliances, WAN optimization appliances, SSD acceleration arrays, public cloud gateways, backup applications, replication applications, etc. so that they all run as a unified stack atop a single shared resource pool.

An Unmanageable Infrastructure – Today’s traditional IT infrastructure has many points of management. Each device has a standalone management interface and often requires expensive specialized training.

Mobility Issues – In traditional IT infrastructure, data is not mobile. VMware has made the server mobile, but the data associated with the virtual machine is still limited in its mobility. Data needs to be disassociated from specialized hardware in order to be as mobile as a virtual machine.

Deficiencies in Responsiveness – The time to data (restore, replicate, clone, etc.), both locally and remotely, is too long. This introduces limitations in terms of required data management and protection best practices.

A Lack of Scalability – Traditional IT scales in very coarse increments. It is often impractical to predict infrastructure requirements, especially multiple years out. Data center managers need a solution that can scale out on demand and as needed while not increasing complexity.

High Cost – Highly functional and highly performing data storage has been dependent on expensive SAN hardware, both from a CapEx (acquisition) and an OpEx (management) stand point.

ESG research confirms these challenges in a spending intentions survey that asked organizations what their most important IT priorities for 2014 were. As shown in Figure 1, the most cited responses include increased use of server virtualization, improved data backup and recovery, managing data growth, and data center consolidation. 1

1 Source: ESG Research Report, 2014 IT Spending Intentions Survey, February 2014.

http://www.esg-global.com/research-reports/2014-it-spending-intentions-survey/



Figure 1. Top Ten Most Important IT Priorities for Organizations in 2014

Source: Enterprise Strategy Group, 2015.

Springpath Data Platform for VMware

The Springpath Data Platform is a scale-out software platform that runs on the same servers that host the hypervisor, VMs, and applications, converging storage and compute functionality onto a single, unified, commodity server-based infrastructure cluster. Founded by early VMware architects who created VMFS and VxLAN, Springpath’s goal was to create software that is differentiated by:

Being a 100% software solution capable of running on existing customer infrastructures using tier-1 vendors including Cisco, Dell, and HP.

Scaling both storage capacity and performance as servers are added to the cluster. Delivering high I/O performance and capacity. Being easy to deploy and use. Providing enterprise-class data services without caveats. Maintaining auto-support monitoring and analytics to help proactively discover, analyze, and address

issues.

22%

23%

23%

23%

23%

24%

25%

29%

32%

32%

0% 5% 10% 15% 20% 25% 30% 35%

Data center consolidation

Business intelligence/data analyticsinitiatives

Major application deployments or upgrades

Regulatory compliance initiatives

Use cloud infrastructure services

Desktop virtualization

Manage data growth

Improve data backup and recovery

Information security initiatives

Increased use of server virtualization

Top 10 most important IT priorities over the next 12 months. (Percent of respondents, N=562, ten responses accepted)



Figure 2. Springpath Data Platform

Standard VMware features such as vMotion, replication, Distributed Resource Scheduler (DRS), and high availability are all fully supported.

Springpath’s hardware agnostic log-structured object (HALO) storage technology seamlessly and automatically leverages separate performance and capacity tiers. Data is fully distributed across disks in all the servers in the cluster to leverage all controller resources and provide high availability. Block checksums and SHA fingerprints assure data integrity. This design enables users to configure a single datastore to house all data on the cluster with no impact on performance.

Logical abstractions of physical resources enable data to be resynchronized and rebalanced when changes occur in the physical state of cluster resources. Built-in space reclamation policies are based on data patterns as written by applications.

Data deduplication is in-line with no measurable overhead. Springpath software performs deduplication at the cache layer to increase cache efficiency and performance, and at the persistence layer to provide space savings. In-line compression, for any workload, also has practically no overhead, providing space savings with uncompromised, sustained, and predictable performance.

Springpath native snapshots and ReadyClones are invoked directly from the Springpath plug-in in the VMware Web Client and have no performance, scalability, or granularity caveats. Springpath snapshots are pointer-based and zero-copy, hence making them space efficient. ReadyClones are customized clones that are space efficient and are also provisioned through the Springpath plug-in in the vSphere client. Both snapshots and clones are integrated into VMware via VAAI.



ESG Lab Validation

ESG Lab performed hands-on evaluation and testing of the Springpath Data Platform at Springpath’s facilities in Sunnyvale, California. Testing was designed to demonstrate the agility, efficiency, and simplicity of the Springpath Data Platform using industry-standard tools and methodologies. ESG Lab was also shown the capabilities of Springpath’s elastic scalability and reliability.

Getting Started

ESG Lab started with a pre-staged test bed as summarized in Figure 3. Two 2U high-density server chassis were employed. Each chassis consisted of four industry-standard x64 servers and 24 2.5” drive slots. Each of the eight servers had five 1 TB SATA hard disk drives and one 480 GB SSD directly attached. Each physical server was redundantly connected to both a 10 gigabit Ethernet switch for data, and a 1 gigabit Ethernet switch for management connectivity.

Figure 3. The ESG Lab Test Bed

ESG Lab Testing

ESG Lab testing began with a review of the configuration of the environment. Each server was running VMware vSphere and Springpath software. The Springpath administration console is accessed by a standard web browser using a DHCP-assigned IP address. The web console provides the ability to manually configure each of the four servers.

For rapid and easy configuration, administrators can create a configuration file in advance, using a web-based utility from the Springpath support website. The configuration file contains all of the pertinent Springpath configuration information, formatted in JavaScript Object Notation (JSON, a lightweight data-interchange format). Using drag-and-drop, the configuration can be quickly applied to the system. An example of the JSON configuration file is shown in the appendix.

Using the web console, ESG Lab accepted the end-user license agreement (EULA), and then clicked on the “Import Configuration” button. This brought up a drag-and-drop box. A JSON configuration file was dragged to the drop box, and the “Next” button was selected. The console then applied the configuration and provided a summary display, as shown in Figure 4.



Figure 4. Initial Configuration Summary

The initial configuration summary shows the IP addresses and host names for each of the four hosts in the chassis, along with other network information and vSphere vCenter Server configuration information.

Once the initial configuration has been completed, all other configuration is performed through VMware’s vSphere management system. While Springpath provides a vSphere plug-in for a few Springpath-specific functions, almost all management is performed directly from the vSphere Web Client.

ESG Lab next started the vSphere Web Client and navigated to the Springpath system in the list of managed systems. Once the system was selected, the Springpath plug-in was used to examine the system. As can be seen in Figure 5, there are four tabs: Getting Started, Summary, Monitor, and Manage.



Figure 5. Springpath Data Platform Management

ESG Lab selected the Manage tab, which by default provides an overview of the Springpath cluster. Administrators are given the option to display information about the hardware appliance or the Springpath datastores. Data can be displayed in tabular form, or as a graphical representation of the server hardware, showing the state of the drive bays with icons indicating solid-state or spinning disks. Hovering the mouse cursor over a specific disk displays a popup window with more detailed information. The graphical display also shows the network connections of the four hosts, including the 1GbE management port and the four 10GbE data ports. Hovering the mouse cursor over a specific Ethernet port shows a popup with more detailed information including the MAC address, device name, and speed of the port.

The tabular format, shown in Figure 5, provides summary state information including the number of hosts and disks, the power supply status, and IP addresses for each host in the server cluster. Administrators can select each individual host to display more detailed information.

After reviewing the state and configuration of both server chassis, ESG selected the “Datastore” button on the Manage tab to create a new datastore (see Figure 6). The top of this view provides a listing of all configured datastores, while the bottom shows the status for the selected datastore. In this case, datastore BMT_1 is mostly free, as indicated by the green section of the pie chart and the table showing the allocated and free space. Also shown is a graph of the IOPS for this datastore over time. The hosts assigned to this datastore can be displayed through the Hosts tab.



Figure 6. Datastore Management

From the Datastore Management tab, administrators can quickly and easily create, modify, and delete datastores, as well as change the datastore to host mappings. ESG Lab clicked on the “Create Datastore” button to create a new datastore. Because Springpath uses wide-striping across all disks in the cluster, administrators do not have to create RAID sets, storage pools, or other complicated storage virtualization schemes. Instead, the only configurable is the size of the datastore. Once the size has been entered, the datastore is created and is almost immediately available for use. It is important to note that while multiple datastores were created during ESG Lab testing, in practice it makes more sense to just create one datastore for the sake of management simplicity.

Next, ESG Lab deployed virtual machines to the newly created datastore. Using vSphere and storage vMotion, an existing gold master VM was moved to the Springpath datastore. As an initial test of the performance of the Springpath file system, the Springpath plug-in in the vSphere Web Client was employed to create 100 clones of the master VM image (see Figure 7). Total elapsed time to create the clones was a fast two minutes and 32 seconds.



Figure 7. Creating 100 Clones of the Golden Master Virtual Machine

After the clone creation, the Summary tab was selected to display summary information on the state of the Springpath Data Platform (see Figure 8). The system reported almost 10TB of provisioned capacity, yet was only consuming 178GB of physical storage space. This translated to a 38% compression rate and 78% deduplication rate, with an overall space savings of greater than 86%.

Figure 8. Springpath Data Platform Summary of State



Springpath provided additional information on the summary page including performance graphs of I/O, bandwidth, and latency for the previous hour. System health was communicated through bar graphs at the top right of the window indicating the amount of storage consumed and the number and health of the hosts and storage controllers.

Why This Matters

To take advantage of the benefits of virtualization and private cloud technology, IT infrastructures have grown complex. Organizations are turning to integrated computing platforms and converged infrastructures as a means of simplifying the environment. ESG research has found that faster deployment times, ease of management, and simplified deployment processes rank among the top six benefits most-cited by surveyed organizations implementing integrated computing platforms.2

ESG Lab validated that the Springpath Data Platform was easy to deploy in less than 15 minutes using the web interface, and easy to manage using the VMware vSphere Web Client. Configuring the system was simplified through the use of automated tools. Using a standard web browser, only a few mouse clicks were required to apply the configuration. Provisioning storage and exporting storage volumes to clients was just as quick and easy using both the vSphere client and the Springpath plug-in. The management interface provides at-a-glance information to show the administrator the current configuration and health of all aspects of the system. With tight vSphere integration, Springpath administrators have a single interface to manage the hardware, Springpath Data Platform, and VMware virtualization, providing a complete, robust, easy-to-use, and agile IT infrastructure.

2 Source: ESG Research Report, Trends in Private Cloud Infrastructure, March 2014.

http://www.esg-global.com/research-reports/trends-in-private-cloud-infrastructure/



High Availability

Springpath Data Platform is designed with enterprise reliability and high availability in mind. Deploying Springpath on highly available servers provides N + 1 fault tolerance for key hardware components. Springpath’s hardware agnostic log-structured object (HALO) storage technology uses wide striping to provide high-performance data protection, ensuring nonstop operation and zero data loss with up to two drive failures across multiple servers in the cluster. Springpath leverages all of the high-availability features built into VMware, including multipathing and VMware HA to provide access to VMs and data even in the event of a Springpath or cluster server failure.

ESG Lab Testing

The next phase of testing focused on demonstrating high-availability operation and recovery during a disk drive failure. While simulated user activity was occurring on VMs in the Springpath Data Platform, ESG Lab faulted a disk drive by manually removing the drive from a server.

ESG Lab observed the status of the server cluster using the VMware vSphere Web Client. After the disk was removed from the server, the Springpath plug-in indicated that the drive had failed. ESG Lab next reinserted the drive into the server. The Springpath plug-in indicated that the drive was now being repaired, as shown in Figure 9. During this sequence, user activity was uninterrupted.

Figure 9. Springpath Disk Status

ESG Lab next simulated a server failure by powering off a server in the cluster using the server’s power switch, without first shutting down the software. ESG Lab observed through the vSphere Web Client that VMware HA detected the server failure and moved that servers’ VMs to other nodes in the cluster.

Next, ESG Lab powered on the server, and manually executed a VM migration to the server. Because the Springpath datastore is shared between all nodes in the cluster, the only step was to use VMware vMotion to move the virtual machine, as shown in Figure 10. VMware Storage vMotion was not required.



Figure 10. Springpath VM Migration

Why This Matters

Data protection of virtualized infrastructure is critical to IT operations. The ability of IT end-users to access information anytime from anywhere is not simply desired; it is expected. Global operations demand 24x7 data access, leaving no window for planned or unplanned downtime. Server and storage consolidation magnify the need for high availability and reliability because a hardware outage will affect many systems and applications, not just one.

Springpath Data Platform is designed to provide flexible tier-1 storage, engineered for efficiency and performance in support of the modern data center. ESG Lab validated that Springpath provided nonstop data services during a hard failure of a disk drive. Springpath leveraged VMware HA services to automatically migrate VMs in the case of a server failure. Because Springpath utilizes all server nodes in the cluster for data storage, VM migration occurred without the need for VMware Storage vMotion, improving the efficiency of high-availability operations.

ESG Lab has confirmed that Springpath Data Platform can provide an always-on, non-disruptive storage environment able to operate through planned maintenance and unplanned faults thanks to a tightly integrated, highly available architecture combined with robust cluster services.



Mixed Workload Performance

Conventional server benchmarks were designed to measure the performance of a single application running on a single operating system inside a single physical computer. Much like traditional server benchmarks, conventional storage system benchmarks were designed to measure the performance of a single storage system running a single application workload. The SPC-1 benchmark, developed and managed by the Storage Performance Council, is a great example. SPC-1 was designed to assess the performance capabilities of a single storage system as it services an online interactive database application.

ESG Lab used a different approach in this validation. Rather than running application-level benchmarks, which stress the CPU and memory of the server, lower level industry-standard benchmarks were used with a goal of measuring the mixed workload capabilities of the Springpath Data Platform running on industry-standard hardware.

ESG Lab’s storage-focused benchmarking uses multiple virtual servers emulating diverse applications, as would be found in a typical data center. The industry-standard Iometer, FIO, and Microsoft Jetstress 2013 utilities were used to generate raw IOPS and emulate the I/O activity of common business-critical application workloads. 3 4 5

ESG Lab Testing

The first test measured raw IOPS:

IOPS: I/O per second, or IOPS, is a measure of the number of operations that a storage system can perform. When a system is able to move a lot of IOPS, it will tend to be able to service more applications and users in parallel. Much like the horsepower rating for a car engine, the IOPS rating for a storage array can be used as an indicator of the power of a storage system engine. A 100% read, 100% random 4KB I/O workload was generated with FIO.

The raw IOPS workload was run first using individually provisioned virtual machines, and then repeated with cloned virtual machines using Springpath’s ReadyClones. Springpath ReadyClone technology provides additional space savings over traditional clones as the ReadyClones are automatically deduplicated. As VMs write data to ReadyClones, the data is both compressed and deduplicated, saving space. Additionally, performance is typically increased due to the higher probability that a deduplicated block will reside in fast cache instead of having to be retrieved directly from disk.

Figure 11. Performance of Full Virtual Machines versus Clones

3 http://sourceforge.net/projects/iometer/ 4 http://www.freecode.com/projects/fio/ 5 http://www.microsoft.com/en-us/download/details.aspx?id=36849

165,280

66,073

164,992

79,318

0

40,000

80,000

120,000

160,000

4KB 100% Read IOPS 4KB 80%Read/20%Write IOPS

IOP

S

Full VMs Clone VMs

http://sourceforge.net/projects/iometer/

http://www.freecode.com/projects/fio/

http://www.microsoft.com/en-us/download/details.aspx?id=36849



As seen in Figure 11, IOPS serviced were nearly identical for both sets of virtual machines. Figure 11 also shows the results of a workload emulating an OLTP database. This workload also used 4KB blocks, with a random mix of 80% read and 20% write operations. For this workload, the cloned VMs were able to sustain 20% more IOPS than the full VMs.

For both the raw read IOPS test and the emulated OLTP database performance test, the Springpath Data Platform was able to maintain virtually the same average response times across full and cloned VMs, with differences of less than 2%.

ESG Lab next compared the performance between VMware snapshots and Springpath native snapshots. VMware snapshots suffer a performance penalty under load and when consolidating snapshots as a result of the built-in redo log. The native snapshots provided by the Springpath Data Platform are optimized for performance.

Again using FIO, ESG Lab reran both the 4KB 100% random read test and the OLTP emulation of 80% reads/20% writes. These tests were executed on a system with the maximum number of VMware snapshots, and then again with the same number of Springpath native snapshots. The results are shown in Figure 12.

Figure 12. Performance of VMware Snapshot versus Springpath Snapshot

Next, industry-standard benchmarks were used to emulate the I/O activity of four common business application workloads:

E-mail: The Microsoft Jetstress 2013 utility was used to generate e-mail traffic. Similar to the Microsoft LoadSim utility used in the VMmark benchmark, Jetstress simulates the activity of typical Microsoft Exchange users as they send and read e-mails, make appointments, and manage to-do lists. The Jetstress utility is, however, a more lightweight utility than LoadSim. Using the underlying Jet Engine database, Jetstress was designed to focus on storage performance.

Database: The Iometer utility was used to generate database-like traffic: read-heavy, small block, 100% random workload response-time sensitive online transaction processing (OLTP).

Web Server: The Iometer utility was used to generate web server traffic. The I/O definition was composed of random reads of various block sizes. The web server Iometer profile used for this test was originally distributed by Intel, the author of Iometer. Iometer has since become an open source project.

58,438

200,903

89,551

217,311

0

50,000

100,000

150,000

200,000

250,000

4KB 80% Read/20% Write 4KB 100% Read

IOP

S

VMware Snapshot Springpath Snapshot



File Services: The Iometer utility was used to generate file server traffic. The I/O definition was composed of random reads and writes of various block sizes. The file server Iometer profile used for this test was originally distributed by Intel, the author of Iometer. Iometer has since become an open source project.

Each of the four workloads was run in parallel in its own dedicated virtual machine. One instance of each workload was set to run against each of the nodes in the cluster. Figure 13 shows results for a four-, six-, and eight-node cluster. It’s important to note that performance was measured in users rather than raw IOPS, and each application will consume a different number of IOPS per user.

Figure 13. Springpath Mixed Workload Performance Scaling

Springpath scaled nearly linearly from just over 54,216 users on a four-node cluster to 96,376 total users on an eight-node cluster. Average response time at the hosts, often used as a measure of application responsiveness, was low, at under 5ms for the four-node cluster. Response times stayed low as the cluster grew and the workload increased, staying far below 20ms, typically used to indicate maximum acceptable response times (indicated by the red dashed line on the chart).

Figure 14 provides more details on the performance results for e-mail, as represented by the Jetstress 2013 tests, for clusters of four, six, and eight nodes.

0

2

4

6

8

10

12

14

16

18

20

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

90,000

100,000

4 Nodes 6 Nodes 8 Nodes

Ave

rage

Res

po

nse

Tim

e (m

s)

Use

rs

E-Mail OLTP Web Access File Services Average Response Time



Figure 14. Springpath Jetstress 2013 Performance

It’s important to note that Jetstress generated much more I/O than was required to support the number of mailboxes reported here while remaining well below the Microsoft recommended threshold of 20ms for average host response time (as indicated by the red dashed line in Figure 14). Table 1 provides the details of the mixed workload testing.

Table 1. Mixed Workload Performance Data

Nodes Exchange 2013

Mailboxes OLTP Users

Web Server Users

File Services Users

Average Read Response Time (ms)

4 10,000 5,673 26,918 11,624 4.72

6 15,000 8,115 37,489 16,943 5.25

8 20,000 9,355 47,770 19,251 6.22

What the Numbers Mean

ESG Lab simulated the traffic of an organization running multiple Exchange 2013 servers, busy OLTP databases, file servers, and web servers supporting thousands of users, all running simultaneously on a Springpath cluster.

A four-node Springpath cluster was able to support 54,216 simulated users, while an eight-node cluster supported 96,376 total simulated users. This represents an excellent scaling factor of 89%, meaning that each node of an eight-node cluster was able to perform 89% of the work of each node of a four-node cluster.

The Springpath cluster yielded remarkably low average response times for each application, and response time stayed low as the workload was increased.

As shown in Figure 13 and Figure 14, all application workloads were well under their respective latency guidelines, which indicates that there was room for either higher performance or more applications to run simultaneously without severely impacting response times.

0

4

8

12

16

20

24

28

32

36

40

44

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

20,000

22,000

4 Nodes 6 Nodes 8 Nodes

Ave

rage

Res

po

nse

Tim

e (m

s)

IOP

S

Mailboxes Jetstress Read Latency (ms)



Why This Matters

Storage scalability and performance are significant challenges for the growing number of organizations embracing server virtualization technology in support of an IT-as-a-service, on-demand delivery model. Storage benchmarks have historically focused on one type of workload (e.g., database or e-mail) and one key performance metric (e.g., IOPS, response time, or throughput). Server benchmarks have typically tested just one server running a CPU-intensive workload that doesn’t stress storage. ESG Lab ran a benchmark designed to assess how real-world applications might behave when running on multiple virtualized servers in a single Springpath cluster.

ESG Lab compared the performance of Springpath ReadyClones with Full VMware VMs, and verified that ReadyClones provided identical or better performance, depending on the workload. Administrators can have full confidence in the performance of the Springpath Data Platform while utilizing ReadyClones to save space through both deduplication and compression. In addition, ESG Lab validated that Springpath’s native snapshot technology provided superior performance to VMware’s snapshot technology, achieving more than 3.4 times more IOPS.

ESG Lab simulated the traffic of multiple Exchange 2013 servers, several busy OLTP databases, file servers, and web servers, all supporting thousands of users and running simultaneously on a single Springpath cluster using a single datastore spanning all hosts. The Springpath cluster scaled non-disruptively as additional nodes, virtual machines, and workloads were brought online, and demonstrated excellent host response time throughout the tests, indicating significant available headroom to grow as performance requirements increase.

ESG Lab has verified that Springpath Data Platform can be deployed to provide cost-effective, easy-to-configure storage for data center environments of all sizes with excellent scalability and performance. Efficient use of a small amount of SSDs for cache enabled the system to support nearly 100,000 total users on a small number of high-capacity drives with excellent response times.



ESG Lab Validation Highlights

ESG Lab validated that Springpath was extremely easy to install and deploy. Using a pre-configured JSON file, a new cluster was deployed in less than 15 minutes. Adding additional nodes to a cluster was non-disruptive and executed in less than a minute using the intuitive web interface.

ESG Lab demonstrated that Springpath simplifies management with no additional training necessary for administrators with working knowledge of the vCenter client. ESG Lab showed that all day-to-day operations could be accomplished using native VMware tools.

ESG Lab validated that Springpath’s proprietary log-structured file system provided always-on deduplication, compression, and optimization of all data, which enhanced system performance, reduced response times, and reduced storage capacity, hence reducing the associated power and space requirements.

ESG Lab validated that data operations performed at a VM level in Springpath were performed more efficiently and quickly than in a traditional virtualized environment, saving valuable time and resources. One hundred clones of a virtual machine completed in a remarkable two minutes, 32 seconds.

ESG Lab demonstrated very nearly linear performance scalability of a mix of common business applications in a virtualized environment. The eight-node Springpath cluster was easily able to support nearly 100,000 users while response times remained manageably low.

Issues to Consider

Springpath technology is server, platform, and application agnostic. Springpath software can be deployed on an array of servers from four major vendors today—Cisco, Dell HP, and Supermicro—and Springpath is in the process of qualifying additional platforms, realizing the flexibility of the technology in terms of scaling performance, capacity, and different ratios of VMs to configurations of CPU, DRAM, SSD, and HDD.

The performance test results presented in this report are based on benchmarks and configurations deployed in a controlled environment. Due to the many variables in each production data center environment, capacity planning and testing in your own environment is recommended.



The Bigger Truth

The ubiquitous adoption of server virtualization requires significant changes to existing storage infrastructure. Storage groups are playing catch up as they design new networked storage solutions to meet the demands that the high I/O and throughput virtualization creates. While virtualization promises to deliver savings in capital expenditures, these savings can be wiped out by the costs of upgrading to the faster storage systems needed to respond to these requirements.

According to ESG research, faster deployment times, ease of management, and simplified deployment processes rank among the top six benefits most-cited by surveyed organizations implementing integrated computing platforms.6

It is clear that consolidation of technologies into a single system minimizes both capital and operational expenditures, but what should also be emphasized is that by storing and maintaining data in a more efficient manner—deduplicated, compressed, and optimized—through its entire lifecycle, organizations will require less hardware and less bandwidth for data transfers, and save valuable administrative time, resulting in greater productivity and faster responses to business needs.

Springpath delivers a software solution that simplifies the deployment of scale-out compute and storage resources while providing agile enterprise-class data management functions and outstanding capacity and performance efficiency via penalty-free in-line deduplication and data compression. Customers looking to build a virtual environment rapidly and easily will find the solution offered by Springpath to be a perfect fit.

ESG Lab was able to confirm extremely rapid and effective implementation of cluster services, snapshots, and cloning in addition to support for multiple VMware functions such as vMotion, Dynamic Resource Scheduler, and HA failover. With the ability to provide all data storage as a single, global datastore, Springpath has changed the playing field for hyperconverged infrastructures, simplifying provisioning and management without compromising performance.

Virtualization will continue to be a disruptive force for IT, requiring new ways of thinking about compute and storage environments as scalability and elasticity become paramount. Springpath is both simple and versatile to deploy and manage with the ability to scale easily for demanding mixed application environments. Springpath’s powerful, proprietary log-structured file system liberates organizations to elastically grow and adapt infrastructure resources to meet rapidly changing business objectives. Infrastructure consolidation has increased the effectiveness of IT resources and decreased complexity, and hyperconverged infrastructure can help organizations take on the next-level challenges of maximizing efficiency while reducing effort and costs. ESG Lab believes that the Springpath Data Platform is strongly positioned to help organizations get to the next level.

6 Source: ESG Research Report, Trends in Private Cloud Infrastructure, March 2014.

http://www.esg-global.com/research-reports/trends-in-private-cloud-infrastructure/



Appendix

Table 2. Test Bed Overview

Software

Springpath Data Platform Version 1.0

Servers

2 – Supermicro 2U Twin2 Server Chassis Four server blades per chassis: CPU (2 per server blade)

2 Intel Xeon E5 2650 @ 2.4 GHz

10 cores per CPU RAM (per server blade)

256 GB DDR4 per server blade Storage (per server blade)

5 7200 RPM 1TB SATA Drives

1 Intel S3500 480 GB SSD

Virtualization Software and Guest Operating Systems

Server Virtualization VMware vSphere ESXi 5.5 Update 2

Guest OS Windows Server 2012 R2 Standard (6.2.9200.0) Characterization Configuration

Characterization Iometer, version 2006.07.27

24 guest VMs (3 per server)

Storage configured as a single datastore in a VMware vSphere environment

E-mail Microsoft Jetstress, version 15.00.0995.000

1 VM per node

Mailboxes: 2,500 per virtual server

IOPS per mailbox: 0.12

Threads: 4

Log buffers: 9,000

Min DB cache: 32 MB

Max DB cache: 256 MB

Insert operations: 40%

Delete operations: 20%

Replace operations: 5%

Read operations: 35%

Lazy commits: 70%

Database Server (OLTP)

Iometer, version 2006.07.27 Four outstanding I/O per physical drive 4KB block size, 100% random, 67% read

File Server Iometer, version 2006.07.27

Four outstanding I/O per physical drive

4KB-64KB block size, 100% random, 80% read

Web Server

Iometer, version 2006.07.27

Four outstanding I/O per physical drive

4KB-256KB block size, 100% random, 100% read



Figure 15. Example Springpath Data Platform JSON Configuration File

{

"clusterName": "benchmark",

"clusterIp": "benchmark-cip.tme.springpathinc.com",

"managementIp": "benchmark-cip-m.tme.springpathinc.com",

"vCenter": {

"url": "tme-vcenter-server.tme.springpathinc.com",

"clustername": "benchmark",

"datacenter": "benchmark"

},

"ipAddresses": {

"10.90.3.51": {

"controller": "benchmark-stctlvm-e.tme.springpathinc.com",

"vMotion": "benchmark-e-v.tme.springpathinc.com",

"hypervisor": "benchmark-e.tme.springpathinc.com",

"ipmi": "benchmark-e-m.tme.springpathinc.com"

},

"10.90.3.53": {

"controller": "benchmark-stctlvm-f.tme.springpathinc.com",

"vMotion": "benchmark-f-v.tme.springpathinc.com",

"hypervisor": "benchmark-f.tme.springpathinc.com",

"ipmi": "benchmark-f-m.tme.springpathinc.com"

},

"10.90.3.55": {

"controller": "benchmark-stctlvm- g.tme.springpathinc.com",

"vMotion": "benchmark-g-v.tme.springpathinc.com",

"hypervisor": "benchmark-g.tme.springpathinc.com",

"ipmi": "benchmark-g-m.tme.springpathinc.com" },

"10.90.3.57": {

"controller": "benchmark-stctlvm-h.tme.springpathinc.com",

"vMotion": "benchmark-h-v.tme.springpathinc.com",

"hypervisor": "benchmark-h.tme.springpathinc.com",

"ipmi": "benchmark-h-m.tme.springpathinc.com"

}

},

"subnetMask": {

"controllerAndvMotion": "255.255.240.0",

"hypervisorAndipmi": "255.255.240.0"

},

"gateway": {

"controllerAndvMotion": "10.90.128.1",

"hypervisorAndipmi": "10.90.0.1"

},

"vlan": {

"controllerAndvMotion": 0,

"hypervisorAndipmi": 0

},

"dnsServers": ["10.90.0.53", "10.64.1.8"],

"ntpServers": ["10.90.0.53"],

"smtpServer": "",

"fromAddress": "",

"timezone": "America/Los_Angeles"

}

20 Asylum Street | Milford, MA 01757 | Tel: 508.482.0188 Fax: 508.482.0218 | www.esg-global.com

http://www.esg-global.com/

Documents

Lab Validation Report - c368768.ssl.cf1.rackcdn.com Validation Report Springpath Data Platform Enterprise Data Management with Cloud Economics and Maximum Simplicity By Tony Palmer,