Inside Microsoft’s Network and Storage VSP/VSC Brian Henry Program Manager Windows Virtualization Microsoft Corporation

Inside Microsoft’s Inside Microsoft’s Network and Storage Network and Storage VSP/VSCVSP/VSC

Brian HenryBrian HenryProgram ManagerProgram ManagerWindows VirtualizationWindows VirtualizationMicrosoft CorporationMicrosoft Corporation

AgendaAgenda

NetworkingNetworkingIntroductionIntroduction

ArchitectureArchitecture

ConfigurationConfiguration

Compatibility RequirementsCompatibility Requirements

Wrap-upWrap-up

StorageStorageRequirementsRequirements


Other topicsOther topics

Virtualized NetworkingVirtualized Networking

IntroductionIntroduction




Wrap-upWrap-up

Networking GoalsNetworking GoalsHigh performanceHigh performance

Avoiding copying dataAvoiding copying data

SecuritySecurityStrong isolationStrong isolation

Extensibility & CompatibilityExtensibility & CompatibilityDifferent topologiesDifferent topologies

Sharing a NICSharing a NIC

Owning a NICOwning a NIC

NATNAT

Interpartition networkingInterpartition networking

Hardware, drivers and standardsHardware, drivers and standards

Networking SolutionNetworking Solution

EmulationEmulationEmulated DEC 21140 support will still be Emulated DEC 21140 support will still be providedprovided

For PXE bootFor PXE boot

For OSes that aren’t enlightenedFor OSes that aren’t enlightened

EnlightenmentEnlightenmentEnlightened networkingEnlightened networking

Much better performanceMuch better performance

Networking devices virtualization-awareNetworking devices virtualization-aware

AgendaAgenda





Wrap-upWrap-up

Networking ComponentsNetworking Components

Virtual network switchVirtual network switchAdmin mostly interacts with thisAdmin mostly interacts with thisLike physical switch, but can add/delete portsLike physical switch, but can add/delete ports

Virtual network cardVirtual network cardGuest-visible componentGuest-visible component

Emulated network cardEmulated network cardProvides emulation where appropriateProvides emulation where appropriateTraps on I/O operationsTraps on I/O operations

Service stackService stackProvides configuration, setup/teardownProvides configuration, setup/teardownProvides WMI interfaceProvides WMI interfaceRuns emulation codeRuns emulation code

Switch FeaturesSwitch Features

Flexible NIC configurationFlexible NIC configurationAdmin can create multiple switchesAdmin can create multiple switchesAdmin decides which switch to use for internal,Admin decides which switch to use for internal,physical, and guest VMs’ Network Interface Cardsphysical, and guest VMs’ Network Interface Cards

Learning algorithmLearning algorithmAnti-denial-of-service implementationAnti-denial-of-service implementation

Port controlPort controlBandwidth controlBandwidth controlVLAN (802.1q)VLAN (802.1q)Monitor modeMonitor mode

Offloading supportOffloading supportLarge Send OffloadLarge Send OffloadChecksum OffloadChecksum Offload

Generic VSP/VSCGeneric VSP/VSC

VSCVSP

VMBus

Network VSP/VSCNetwork VSP/VSCParent Partition Child Partition

Child OSChild OSParent OSParent OS

NetworkVSC

NetworkVSP

VMBus

Physical NICPhysical NIC

Virtual MachineVirtual MachineNICNIC

InternalInternalNICNIC

Virtual SwitchVirtual Switch

NDIS Interaction With VSCNDIS Interaction With VSC

NDIS

Network VSC(NDIS Miniport

Driver)

Child PartitionChild Partition

NDIS Interaction With VSPNDIS Interaction With VSPParent PartitionParent Partition

Vendor-ProvidedVendor-ProvidedPhysical NIC DriverPhysical NIC Driver

(NDIS Miniport Driver)(NDIS Miniport Driver)

Network VSP(NDIS IM Driver)

NDIS

Physical NICPhysical NIC


AgendaAgenda





Wrap-upWrap-up

Switch ConfigurationSwitch Configuration

Per portPer portMonitor modeMonitor mode

VLAN IDVLAN ID

Bandwidth controlsBandwidth controls

Per switchPer switchSwitch nameSwitch name

Configuration is always through Configuration is always through virtualization stackvirtualization stack

Not SNMP, not TelnetNot SNMP, not Telnet

Configuration SampleConfiguration SampleStandardStandard


PhysicalPhysicalNICNIC

Child OSChild OS(1) Stack(1) Stack

Parent OS StackParent OS Stack


Configuration SampleConfiguration SampleDual physical NICsDual physical NICs


PhysicalPhysicalNIC #1NIC #1




PhysicalPhysicalNIC #2NIC #2

Configuration SampleConfiguration SampleChild Partition IsolationChild Partition Isolation






Virtual SwitchVirtual Switch Child OSChild OS(3) Stack(3) Stack


Configuration SampleConfiguration SampleDedicated NICDedicated NIC






Virtual SwitchVirtual Switch Child OSChild OS(3) Stack(3) Stack


Physical NIC ConfigurationPhysical NIC Configuration

MAC addressMAC address

Virtual switch port bindingVirtual switch port binding

Virtual NIC ConfigurationVirtual NIC Configuration

MAC addressMAC addressDynamicDynamicStaticStatic

Virtual switch port bindingVirtual switch port binding

Microsoft has new IEEE Organizationally Microsoft has new IEEE Organizationally Unique Identifier (OUI) for virtual NICsUnique Identifier (OUI) for virtual NICs

OUI is 00-15-5DOUI is 00-15-5DUsed when dynamically allocated Used when dynamically allocated Avoids collisions with earlier virtualization Avoids collisions with earlier virtualization effortsefforts

MAC Address ConflictsMAC Address Conflicts

Networking infrastructure is not Networking infrastructure is not responsible for MAC address conflict responsible for MAC address conflict detection or resolutiondetection or resolution

Microsoft Windows virtualization in-box Microsoft Windows virtualization in-box solution will prevent conflicts within scope of solution will prevent conflicts within scope of physical machine (best effort)physical machine (best effort)

External management solutions are required External management solutions are required to prevent conflicts outside scope of physical to prevent conflicts outside scope of physical machinemachine

Configuration MechanismsConfiguration Mechanisms

UIUIBasic out-of-box configurationBasic out-of-box configuration

Built on top of WMI, no private APIBuilt on top of WMI, no private API

WMIWMIFully documented Fully documented

Basis for command-line supportBasis for command-line support

Basis for external management appsBasis for external management apps

WMI Details For NetworkingWMI Details For Networking

Objects exposed: Switches, ports,Objects exposed: Switches, ports,and NICsand NICs

Create switches, ports and NICs,Create switches, ports and NICs,bind to switchesbind to switches

Retrieve statistical/accounting dataRetrieve statistical/accounting dataPackets sentPackets sent

Packets receivedPackets received

Packets droppedPackets dropped

Per port and per switch Per port and per switch

WMI Object Hierarchy - SwitchWMI Object Hierarchy - Switch

CIM_SwitchPort CIM_LANEndpointCIM_SwitchService

CIM_ForwardingService

CIM_NetworkService

CIM_Service

1

0..*

Msvm_SwitchesAmong

CIM_ProtocolEndpoint

CIM_ServiceAccessPoint

CIM_EnabledLogicalElement

1 1

Msvm_SwitchPortConnection

CIM_TransparentBridgingService

CIM_LogicalElement

CIM_DynamicForwardingEntry

1

*

Msvm_TransparentBridgingDynamicForwarding

1..* *Msvm_SwitchServiceTransparentBridging

1..*

*

Msvm_SwitchPortDynamicForwarding

Msvm_SwitchService Msvm_TransparentBridgingService Msvm_SwitchPort Msvm_EthernetCable

Msvm_DynamicForwardingEntry

WMI Object Hierarchy - VDEVWMI Object Hierarchy - VDEV

CIM_NetworkPort

CIM_LogicalPort

CIM_LogicalDevice

1

1

Msvm_EthernetPortCableCIM_EthernetPort

CIM_ComputerSystem

1

1

Msvm_VirtualComputerSystemEthernetCable

VDEV Classes

Msvm_EthernetCable

Msvm_EmulatedEthernetPort

Msvm_SyntheticEthernetPort

Switch Classes

WMI Object Hierarchy – WMI Object Hierarchy – Resource Allocation Resource Allocation

CIM_NetworkPort

CIM_LogicalPort

CIM_LogicalDevice

CIM_EthernetPort

Msvm_EmulatedEthernetPort Msvm_SyntheticEthernetPort

CIM_ManagedElement

CIM_SettingData

CIM_ResourceAllocationSettingData

Msvm_EmulatedEthernetPortSettingData Msvm_SyntheticEthernetPortSettingData

CIM_ManagedSystemElement

CIM_LogicalElement

CIM_ResourcePool

Msvm_EmulatedEthernetPortResourcePool Msvm_SyntheticEthernetPortResourcePool

CIM_EnabledLogicalElement

1

1

Msvm_ElementSettingData

1

1

Msvm_ElementSettingData

1

1

Msvm_AllocatedFromPool

1

1

Msvm_AllocatedFromPool

Config State And MigrationConfig State And Migration

Config state is kept in XML,Config state is kept in XML,which is private to the VDevwhich is private to the VDev

Initial config is achieved through WMIInitial config is achieved through WMI

Config changes made through WMIConfig changes made through WMI

XML schema will not be publishedXML schema will not be published

XML is used for serialization/XML is used for serialization/deserialization on store and startupdeserialization on store and startup

We have made the assumption thatWe have made the assumption thatmigration does not depend on statemigration does not depend on state

Policy must be defined above networkingPolicy must be defined above networkingstack by management componentsstack by management components

AgendaAgenda





Wrap-upWrap-up


NDIS driverNDIS driver

Listen in promiscuous modeListen in promiscuous modeReceive packets with any destination MAC addressReceive packets with any destination MAC address

Accept the OID_GEN_CURRENT_PACKET_FILTERAccept the OID_GEN_CURRENT_PACKET_FILTERwith a value of NDIS_PACKET_TYPE_PROMISCUOUS with a value of NDIS_PACKET_TYPE_PROMISCUOUS

Flexible source MACFlexible source MACTransmit packets with any source MAC addressTransmit packets with any source MAC address

802.3802.3Ethernet framesEthernet frames

Return NdisMedium802_3 when handling Return NdisMedium802_3 when handling OID_GEN_MEDIA_SUPPORTEDOID_GEN_MEDIA_SUPPORTED

Note: 802.11 is not compatible and not supportedNote: 802.11 is not compatible and not supported

Compatibility RequirementsCompatibility Requirements(future)(future)

Use PCI bus architecture to enable device Use PCI bus architecture to enable device assignmentassignment

Each device should have its own PCI Bus-Each device should have its own PCI Bus-Device-Function (BDF)Device-Function (BDF)

AgendaAgenda





Wrap-upWrap-up

Your Input Is NeededYour Input Is Needed

We want your input to help prioritize our We want your input to help prioritize our efforts for future versions of the virtual efforts for future versions of the virtual environmentsenvironments

Device assignmentDevice assignment

Chimney (TOE) supportChimney (TOE) support

SNMP agent for switchSNMP agent for switch

MAC address listsMAC address lists

Virtualized StorageVirtualized StorageGoalsGoals

Participants will leave withParticipants will leave withan understanding ofan understanding of

How Microsoft intends to enableHow Microsoft intends to enableefficient storage device virtualizationefficient storage device virtualization

What options are available forWhat options are available forbacking storagebacking storage

How common enterprise scenarios workHow common enterprise scenarios work

Agenda (Storage)Agenda (Storage)

RequirementsRequirements

ArchitectureArchitectureDevice emulationDevice emulation

Enlightened I/OEnlightened I/O

Other topicsOther topicsExtensibilityExtensibility

VHD utilityVHD utility

ClusteringClustering

VSSVSS

RequirementsRequirementsFunctional goalsFunctional goals

Provide a general purpose, high-Provide a general purpose, high-speed virtual storage infrastructurespeed virtual storage infrastructure

Allow format-independenceAllow format-independencebetween the backing storebetween the backing storeand the child partitionand the child partition

Using an image, either ISO or VHDUsing an image, either ISO or VHD

Using an entire physical driveUsing an entire physical drive

Using NAS or SANUsing NAS or SAN

Solution RequirementsSolution RequirementsPerformancePerformance

Provide greatly improved performance for Provide greatly improved performance for identified scenariosidentified scenarios

Minimize code path lengthMinimize code path length

AgendaAgenda







VSSVSS

Device EmulationDevice EmulationComponentsComponents

SCSI emulatorSCSI emulatorEmulates Adaptec 7870 SCSI controllerEmulates Adaptec 7870 SCSI controller

IDE emulatorIDE emulatorEmulates Intel PIIX4 IDE controllerEmulates Intel PIIX4 IDE controller

DriversDriversUses native in-box drivers for WindowsUses native in-box drivers for Windows®®

Uses drivers common in other OSesUses drivers common in other OSes

Device EmulationDevice EmulationArchitectureArchitecture

Parent Child

User-Mode User-Mode

Kernel-ModeKernel-Mode

Image Parser

AdaptecMiniport

Storport

Hypervisor

Virtual Storage Server (VSP)

I/O Handlers

IDE Controller

SCSI Controller

IDE Disk

IDE DVD

SCSI Disk

Windows Virtualization Stack Worker Process

Volume

Disk

Partition

FileSystem

File System

StorPortMiniport

Disk

Partition

Volume

Virtualization Stack

I/O operationsI/O operationscause trapscause traps

Hypervisor intercepts Hypervisor intercepts and redirectsand redirects

Emulations make Emulations make requests ofrequests ofstorage serverstorage server

Storage serverStorage serverpasses requestspasses requestson to a VHD parser on to a VHD parser

Lots of context Lots of context switchingswitching

Device EmulationDevice EmulationBenefitsBenefits

ProsProsStandard set of supported legacy devicesStandard set of supported legacy devices

Adequate for non-performance-intensive Adequate for non-performance-intensive applicationsapplications

ConsConsInadequate performance for some Inadequate performance for some applicationsapplications

Device EmulationDevice EmulationWhen to useWhen to use

During bootDuring boot

When using unenlightenedWhen using unenlightenedoperating systemsoperating systems

Enlightened I/OEnlightened I/OComponentsComponents

Storage VDevStorage VDevUsed to instantiate, configure, andUsed to instantiate, configure, andcontrol enlightened storage devicescontrol enlightened storage devices

Exposes WMI interfaceExposes WMI interface

Virtual storage serverVirtual storage serverUses SRB protocol to communicateUses SRB protocol to communicatewith clientwith client

Top edge interfaces with storage VDevTop edge interfaces with storage VDev


Image parsersImage parsersTranslates storage requests and responses Translates storage requests and responses from SRB to backing store formatfrom SRB to backing store format

ParsersParsersVHDVHD

ISO (read-only)ISO (read-only)

Pass-throughPass-through


Virtual storage miniportVirtual storage miniportStorport miniportStorport miniport

VMBusVMBusInter-partition communication mechanismInter-partition communication mechanism

Enlightened I/OEnlightened I/OArchitectureArchitecture

Parent Child

User Mode

Kernel Mode

Viridian Virtualization Stack Worker Process

Image Parser

VMBUS

Virtual Storage Miniport (VSC)

StorPort

VM

SR

Bs

Virtual Storage Server (VSP)

Hardware

Fast Path Filter

VM

S

RB

s

User Mode

Kernel Mode

Volume

StorPortMiniport

Disk

Partition

File System

Disk

Volume

Partition

File System

No I/O trapsNo I/O trapsLittle hypervisor Little hypervisor InvolvementInvolvementEnlightenedEnlightenedrequests makerequests makerequests ofrequests ofstorage serverstorage serverStorage serverStorage serverpasses on requestspasses on requests

VHD parserVHD parserDirectly to a LUNDirectly to a LUN

Much less context Much less context switchingswitching

Enlightened I/OEnlightened I/OSupported backing storageSupported backing storage

Direct-attached storageDirect-attached storageATAATA

SCSISCSI

OtherOther

Network-attached storageNetwork-attached storage

Storage area networksStorage area networks

Enlightened I/OEnlightened I/OBenefitsBenefits

ProsProsMuch better performance than emulationMuch better performance than emulation

ConsConsDoes not work for unenlightened guests Does not work for unenlightened guests

Enlightened I/OEnlightened I/OWhen to useWhen to use

Wherever possibleWherever possibleWherever VSCs are availableWherever VSCs are available

AgendaAgenda







VSSVSS

Other TopicsOther TopicsExtensibilityExtensibility

Image parser not extensible in Image parser not extensible in first releasefirst release

Third parties are encouraged to add value Third parties are encouraged to add value for specialized marketsfor specialized markets

Other TopicsOther TopicsVHD utility functionsVHD utility functions

CreationCreationCreating new disksCreating new disks

CompactionCompactionCompacting dynamic disksCompacting dynamic disks

ExpansionExpansionExpanding dynamic disks beyondExpanding dynamic disks beyondoriginal max sizeoriginal max size

Merging and forkingMerging and forkingMerging differencing disksMerging differencing disks

Inserting new differencing disksInserting new differencing disks

Other TopicsOther TopicsClustering and VSSClustering and VSS

Clustering “just works”Clustering “just works”Host-side or guest-sideHost-side or guest-side

Volume Shadowing Services “just works”Volume Shadowing Services “just works”Child receives VSS messages from parent Child receives VSS messages from parent (for enlightened guests)(for enlightened guests)

Call To ActionCall To Action

Be aware of compatibility requirementsBe aware of compatibility requirements

Consider your value addConsider your value add

Send us your feedback at Send us your feedback at [email protected]@microsoft.com

mailto:[email protected]

Questions?Questions?

© 2006 Microsoft Corporation. All rights reserved. Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries.The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions,

it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.

Documents

Inside Microsoft’s Network and Storage VSP/VSC Brian Henry Program Manager Windows Virtualization Microsoft Corporation