vmware horizon with view and virtual san reference ... · this reference architecture showcases...

VMware Horizon with View and Virtual SAN Reference Architecture VMware Horizon with View T E C H N I C A L W H I T E PA P E R

VMware Horizon with View and Virtual SAN Reference Architecture

T E C H N I C A L W H I T E PA P E R / 2

Table of Contents

Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

VMware Reference Architecture Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Hardware Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6

VMware vSphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

VMware Virtual SAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

Horizon with View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

VMware View Storage Accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9

Test Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Login VSI 4.1 Workload Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Test 1: 400 Medium-Workload Linked-Clone Desktops . . . . . . . . . . . . . . . . . . . . . . . . 10

Test 2: 400 Heavy-Workload Linked-Clone Desktops . . . . . . . . . . . . . . . . . . . . . . . . . . 13

VMware View Planner Workload Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

View Operations Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Provisioning 400 Linked-Clone Desktops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Refreshing 400 Linked-Clone Desktops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Recomposing 400 Linked-Clone Desktops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Deleting a Pool of 400 Linked-Clone Desktops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Powering On 400 Desktops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Resiliency Test: One-Node Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

System Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

vSphere Clusters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

ESXi Servers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Virtual SAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Virtual SAN Storage Policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Networking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

Horizon with View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

View Global Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

VMware View Manager Global Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

vCenter Server Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

View Manager Pool Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Test Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Login VSI 4.1 Workload Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Medium Workload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Heavy Workload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39



View Planner 3.0 Workload Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Virtual Machine Test Image Build . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

System Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Hosts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Virtual SAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Disk Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Objects and Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Management Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Bill of Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

About the Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51



Executive SummaryThis reference architecture showcases VMware® Horizon™ with View™ running on VMware Virtual SAN™ and is based on real-world test scenarios, user workloads, and infrastructure system configurations. It uses Dell R720 PowerEdge rack mount servers with local storage to support a scalable and cost-effective VMware Horizon with View linked-clone desktop deployment on VMware vSphere® 5.5.

Extensive user experience and operations testing, including Login VSI desktop performance and VMware View Planner testing on 400 desktops, revealed world-class performance at an extremely low cost. VMware Virtual SAN technology allows easy scalability while maintaining superior performance at a competitive price point.

The results are summarized here and described later in this paper.

400 Linked ClonesDeployed

55 Minutes

66 Minutes

5 Minutes

34 Minutes

5 Minutes

12,019 IOPS, 3 millisecond

latency

VSImax v4: Not reached

400 Linked ClonesRecomposed

400 linked clones started

400 linked clonesrefreshed

Single-host failure, recovery time objective for 77 desktops

Login VSI medium workload400 linked clones, 100% concurrency

VSImax v4: 399 desktops

Login VSI heavy workload400 linked clones, 100% concurrency

View Planner: 400 desktops

View Planner medium workload400 linked clones, 100% concurrency

Peak IOPS per host 100 concurrent desktops, 3 millisecond average latency

400

400

400

400

Figure 1. Test Results

Outstanding Desktop Performance

•Performancesimilartohigh-endphysical PCs

•Excellentuserexperiencecharacteristics, even with diverse use cases and burst- heavy workloads

Fast Desktop Operations

•Userdesktoppoolsprovisionedquickly

•Enterpriseusecasesupandrunning the same day as requested

•Newdesktops,applications,andhot fixes rolled out within a single maintenance window



VMware Reference Architecture OverviewVMware reference architectures are built and validated by VMware and supporting partners. They are designed to address common use cases; examples include enterprise desktop replacement, remote access, business process outsourcing, and disaster recovery. A reference architecture describes in detail the environment and workload used to simulate realistic usage, and draws conclusions based on that particular reference architecture.

This guide is intended to help customers—IT architects, consultants, and administrators—involved in the early phases of planning, design, and deployment of Horizon with View–based solutions. The purpose is to provide a standard, repeatable, and highly scalable design that can be easily adapted to specific environments and customer requirements.

The reference architecture “building block” approach uses common components to minimize support costs and deployment risks during the planning of large-scale, Horizon with View–based deployments. The building block approach is based on information and experiences from some of the largest VMware deployments in production today. While drawing on existing best practices and deployment guides pertinent to many of the individual specific components, the reference architectures are tested and validated in the field and described in detail.

Some key features that can help an organization get started quickly with a solution that integrates easily into existing IT processes and procedures include:

•Standardized,validated,repeatablecomponents

•Scalabledesignsthatallowroomforfuturegrowth

•Validatedandtesteddesignsthatreduceimplementationandoperationalrisks

•Quickimplementation,reducedcosts,andminimizedrisk



OverviewThis reference architecture uses common components to minimize support costs and deployment risks.

The desktop virtualization solution, which combines the best of breed of data center, virtualization, and network technologies, uses Dell R720 rack mount servers with local solid state drives (SSD) and hard disk drives (HDD) running on the vSphere 5.5 software suite for desktop workloads. In addition, standard rack mount servers running vSphere are used for server workloads. The View 5.3.2 environment runs Windows 7 virtual desktops provisioned by VMware View Composer™.

The Virtual SAN storage platform for desktop workloads allows the solution to scale linearly, with each host capable of supporting approximately 100 users per host. This reference architecture shows 400 desktops running on four VMware ESXi™ hosts. VMware testing, however, has shown that View with Virtual SAN can scale up to 1,600 desktops on 16 hosts in a single cluster.

Hardware Components

Virtual SAN Cluster:4 Dell PowerEdge R720

Dell Force10 10GbE Ethernet Switches

3 Infrastructure Hosts

Figure 2. Hardware Infrastructure

For server workloads, the solution uses existing virtual infrastructure hosts using three standard rack mount servers.

Desktop workloads use Dell R720 rack mount servers, which offer high-density memory, balanced I/O, and the latest processors for enterprise virtualization and business-processing environments. The system is optimized for running in virtualized and cloud-computing environments.

The Dell Networking S-Series S4810 is an ultra-low-latency 10/40 Gigabit Ethernet (GbE) top-of-rack switch that is purpose-built for applications in high-performance data center and computing environments. Its compactdesignprovides48dual-speed1/10GbE(SFP+)portsandfour40GbEQSFP+uplinkstoconserverackspace. The switches provide 10GbE network connectivity for management, Virtual SAN, and desktop traffic.

Local SSDs and HDDs are used in conjunction with Virtual SAN technology to provide a scalable and enterprise-class storage solution. Each ESXi host has a disk group consisting of one SSD and four HDDs. The disk groups are combined to form a Virtual SAN datastore. This next-generation storage platform combines powerful and flexible hardware with advanced efficiency, management, and software-defined storage.



VMware vSphere

vSphere is the industry-leading virtualization platform for building cloud infrastructures. It enables users to run business-critical applications with confidence and respond quickly to business needs. vSphere accelerates the shift to cloud computing for existing data centers and underpins compatible public cloud offerings, forming the foundation for the industry’s best hybrid cloud model.

VMware Virtual SAN

Virtual SAN is a hypervisor-converged, software-defined storage platform that is fully integrated with vSphere. Virtual SAN aggregates locally attached disks of hosts that are members of a vSphere cluster to create a distributed shared storage solution. Because Virtual SAN sits directly in the I/O data path, it can deliver the highestlevelsofperformance,scalability,andresiliencewithouttaxingtheCPUwithadditionaloverhead.Virtual SAN enables the rapid provisioning of storage within VMware vCenter™ during virtual machine creation and deployment operations.

Virtual SAN

vSphere

VM VM VM VM

Virtual SAN Clustered Datastore

Figure 3. Virtual SAN Clustered Datastore

Virtual SAN uses a hybrid disk architecture that leverages flash-based devices for performance and magnetic disks for capacity and persistent data storage. Its distributed datastore is an object-store file system that leverages the vSphere Storage Policy-Based Management feature to deliver centrally managed, application-centric storage services and capabilities. Administrators can specify storage attributes, such as capacity, performance, and availability, as a policy on a per virtual machine basis. The policies dynamically self-tune and load-balance the system so that each virtual machine has the right level of resources.

Horizon with View

Horizon with View brings the agility of cloud computing to the desktop by transforming desktops into highly available and agile services delivered from your cloud. View delivers virtual sessions that follow end users across devices and locations. It enables fast, secure access to corporate data across a wide range of devices, including Mac OS, Windows, and Linux machines and iOS and Android tablets.



You can use View with VMware vCenter Server™ to create desktops from virtual machines that are running on ESXi hosts and to deploy these desktops to end users. After you create a desktop, authorized end users can use Web-based or locally installed client software to connect securely to centralized virtual desktops, back-end physical systems, or terminal servers. View uses your existing Active Directory infrastructure for user authentication and management.

Figure 4. Horizon with View Components



VMware View Storage Accelerator

All tests performed in this reference architecture used View Storage Accelerator, an in-memory host caching capability that uses the Content-Based Read Cache (CBRC) feature in ESXi hosts. CBRC provides a per host RAM-based solution for View desktops, considerably reducing the read I/O requests that are issued to the storage layer. It also addresses boot storms—when multiple virtual desktops are booted at once—which cause a large number of reads. CBRC is beneficial when administrators or users load applications or data frequently.

View Storage Accelerator minimizes TCO in View deployments by reducing peak IOPS by 80 percent and peak throughput up to 65 percent.

T E C H N I C A L W H I T E PA P E R / 1 0


Test ResultsThe test results are summarized in the following sections. The test methodology is described later in this paper.

Login VSI 4.1 Workload Testing

The test used Login VSI 4.1 to load the system with simulated desktop workloads using common applications like Microsoft Office, Internet Explorer, and Adobe Reader.

We ran two tests on 400 desktops to observe Virtual SAN performance: One test applied a medium workload and the other a heavy workload, both with 100 percent concurrency.

Test 1: 400 Medium-Workload Linked-Clone DesktopsInTest1,theaveragehostCPUusagereachedabove95percentonallESXihostsat400desktopsundermediumworkloadwith100percentconcurrency.DespitehighCPUusage,VSImaxv4wasnotreached.

Figure 5. VSImax Not Reached on Login VSI Medium Workload, 400 Desktops



Figure 6. ESXi CPU Usage During Login VSI Medium Workload

ThefollowinggraphcomparestheaverageVirtualSANIOPSperESXiservertotheaverageVirtualSANCPUusage(MHz)perESXiserver.ESXiVirtualSANCPUusageaveraged2.5GHzandpeakedatjustover5GHzduring an I/O-intensive phase of the workload that had over 11,000 IOPS.

Figure 7. ESXi Virtual SAN IOPS Versus ESXi Virtual SAN CPU Usage



The ESXi Virtual SAN latency averaged 3 milliseconds (ms) and peaked at just over 7ms during an I/O-intensive phase that reached over 11,000 IOPS.

Figure 8. ESXi Virtual SAN IOPS Versus ESXi Virtual SAN Latency

Highlights of Test 1:

•VSImaxv4notreachedatthebaselineof1,486

•CPUusagehigh,butmemoryusageremainedbelow75percent

•ExcellentVirtualSANclientresponsetimes,evenathighload(average3mslatency,peak7msatover11,000IOPS per ESXi host)

•Peakof47,045IOPSonVirtualSANdatastore(51percentwrites,49percentreads)

CPU Usage 46,608MHz

173.55GB

XMIT 45,625 / RCV 43,600

11,761 IOPS / 128,628KBps

Memory Usage

Network Adapter

Storage Adapter

Figure 9. ESXi Host Metrics – Test 1



Test 2: 400 Heavy-Workload Linked-Clone DesktopsInTest2,theCPUwassaturatedat100percentusageacrossallESXihostsunderheavyworkloadwith100percentconcurrency.VSImaxv4399atbaselineof1470.

Figure 10. VSImax 399 During Login VSI Heavy Workload, 400 Desktops

Figure 11. ESXi CPU Usage During Login VSI Heavy Workload



VirtualSANCPUusageperESXihostaveraged4.5GHzandpeakedatjustover7GHzduringanI/O-intensivephase of over 12,000 IOPS per ESXi host.

Figure 12. ESXi Virtual SAN IOPS Versus ESXi Virtual SAN CPU Usage (MHz)

The ESXi Virtual SAN latency averaged 2.7ms and peaked at just over 7ms.

Figure 13. ESXi Average Virtual SAN IOPS Versus Virtual SAN Latency



Highlights of Test 2:

•VSImaxv4399atthebaselineof1470

•CPUusagehigh,butmemoryusageremainedbelow75percent

•ExcellentVirtualSANclientresponsetimes,evenathighload(average2.7ms,peak7mslatencyatover12,000IOPS per ESXi host)


CPU Usage 51,140MHz

185GB

XMIT 49,476 / RCV 50,288KBps

12,019 IOPS / 128,203KBps

Memory Usage

Network Adapter

Storage Adapter

Figure 14. ESXi Host Metrics – Test 2



VMware View Planner Workload Testing

View Planner simulates user workloads by randomly running operations of applications typically used in a Windows desktop environment. This test checked the operational latencies of the 400 linked-clone desktops at 100 percent concurrency.

Figure 15. View Planner Operational Latencies for 400 Virtual Machines

In the following table, for a list of the operations performed in Group A and Group B, see the Test Methodology section.

TEST GROUP OPERATION TYPE RESULT

Group A Interactive or fast-running operations that areCPUbound,suchasbrowsingthrougha PDF file or modifying a Word document.

95thpercentile:0.528723s(BR:<=1.0sec)

Group B Long-running slow operations that are I/O bound, such as opening a large document or saving a PowerPoint file.

95thpercentile:3.249898s(BR:<=6.0sec)

Table 1. View Planner Quality of Service for 400 Virtual Machines with a Think Time of 5 Seconds



Figure 16. View Planner Medium Workload – CPU Usage over Time

The ESXi Virtual SAN latency averaged 2.6ms and peaked at over 4ms at 5,337 IOPS.

Figure 17. ESXi Virtual SAN IOPS Versus ESXi Virtual SAN Latency



Highlights of View Planner Workload Test:

•ViewPlannerGroupA(0.528723s<=1.0)andGroupB(3.249898s<=6.0)resultsarecomfortablywithinthresholds

•CPUusageaveraged85percentduringworkload,andmemoryusageremainedbelow75percent

•ExcellentVirtualSANclientresponsetimesoccurred,withanaverageof2.6msandapeakof4.7ms


CPU Usage 45,604MHz

177.33GB

XMIT 20,001 / RCV 16,795KBps

9,828 IOPS / 85,454KBps

Memory Usage

Network Adapter

Storage Adapter

Figure 18. ESXi Host Metrics – View Planner



View Operations Tests

Provisioning 400 Linked-Clone DesktopsIn this test, a new pool of 400 linked-clone virtual desktops is provisioned on the Virtual SAN datastore, with about 100 per ESXi host. To complete this task, View Composer creates a replica copy of the 24GB base image (9.42GBthin)ontheVirtualSANdatastore.ViewComposercreatesandcustomizesthedesktopsandjoinsthem to the Active Directory domain. It then takes a snapshot of the virtual desktop, and the desktop goes into an available state.

It took less than 55 minutes to provision 400 Windows 7 linked-clone virtual desktops and for them to appear in the available state in the View Administrator console.

Figure 19. View Provisioning Operation – 400 Linked-Clone Desktops



Refreshing 400 Linked-Clone DesktopsIn a refresh operation, a virtual desktop reverts to its snapshot. The OS disk of each virtual desktop is restored to its original state and disk size.

It took 34 minutes to refresh 400 Windows 7 linked-clone virtual desktops to their original base image.

Figure 20. View Refresh Operation – 400 Linked-Clone Desktops



Recomposing 400 Linked-Clone DesktopsIn a recompose operation, a virtual desktop OS disk is changed to a new base image and snapshot. This feature allows administrators to push out patches and software updates with ease.

In this operation, View Composer creates a replica of the new base image on the Virtual SAN datastore, creates a new OS disk for each virtual desktop, and deletes the old one. The new desktop is then customized, and a new snapshot created.

It took 66 minutes to recompose 400 Windows 7 linked-clone virtual desktops to a fresh base image.

Figure 21. View Recompose Operation – 400 Linked-Clone Desktops



Deleting a Pool of 400 Linked-Clone DesktopsThis test deleted a desktop pool, destroying the associated virtual desktops and replicas. Deleting a pool of 400 linked-clonevirtualdesktopstook29minutes.

Figure 22. View Pool Deletion Operation – 400 Linked-Clone Desktops

Powering On 400 DesktopsThe power-on test was carried out on a four-node Virtual SAN cluster in vCenter. It took just under five minutes for all the virtual desktops to be ready for user login.

Figure 23. View Power-On Operation – 400 Linked-Clone Desktops



Resiliency Test: One-Node Failure

A single Virtual SAN node hardware failure was simulated for a cluster with four hosts and 300 running virtual desktops, all under simulated workload.

Figure 24. Resiliency Testing – Recovery Time for a Single-Node Failure

An ESXi host with 77 running virtual desktops was reset, and all the virtual machines became unavailable. VMware vSphere High Availability restarted the virtual desktops on the other Virtual SAN cluster nodes, and all were ready for user login within five minutes of the simulated failure.

The power was restored to the Virtual SAN node some minutes later. The node rejoined the Virtual SAN cluster, and VMware vSphere Storage DRS™ rebalanced the load across all ESXi hosts in the cluster.

Note: A single-node failure does not trigger an immediate rebuild after a host failure is detected. If a failure that returns an I/O error is detected, such as a magnetic disk or SSD, Virtual SAN immediately responds by rebuilding the disk object. However, for host failures that do not return an I/O error, Virtual SAN has a configurable repair delay time (60 minutes by default) after which components are rebuilt across the cluster. Virtual SAN prioritizes the current workload over rebuilding to minimize the impact on cluster performance.



System ConfigurationsThe following sections describe how the reference architecture components were configured.

Architecture

VMware Virtual SAN integrates with the View pod and block design methodology, which comprises the following components.

View Connection Server – A View Connection Server supports up to 2,000 concurrent connections. Our testing consisted of one View Connection Server, but for production use, a minimum of two is recommended, operating in active/active mode. The two View Connection Servers actively broker and possibly tunnel connections.

View block – View provisions and manages desktops through vCenter. Each vCenter instance supports up to 10,000 virtual desktops. The tests used one vCenter and one Virtual SAN cluster with four hosts. Virtual SAN supports 100 virtual machines per host and clusters of up to 32 hosts—up to 3,200 desktops per Virtual SAN cluster.

Note: The maximum HA-protected virtual machines in a vSphere cluster is 2,046 per datastore.

Desktop Cluster

Ethernet Network

vCenter Server

View Block

View Management Block

View Connection Server

Virtual SAN Datastore

View Desktop Pool

Figure 25. View Pod Configuration

View management block – A separate vSphere cluster was used for management servers to isolate the volatile desktop workload from static server workload. Because our testing used only one vCenter, it incorporated the management and View blocks, but for larger deployments, it is recommended to have a dedicated vCenter for the management and View blocks.



vSphere Clusters

A four-node Virtual SAN cluster was deployed to support 400 virtual desktops. Each Dell R720 server had an identical configuration and ESXi booted from the local SD card.

For the management virtual machines, a three-node cluster was deployed on identical servers.

Virtual SAN DatastoreManagement Datastore

Manage

Management Cluster Desktop Cluster

SQL Server

VM VM VM VM

vCenter/View

Composer

View ConnectionServer

VM VM VM VM

ActiveDirectory

VM VM

vCenterOperations

VMware vCenter Server

Figure 26. vSphere Cluster Design



The Virtual SAN cluster had the following settings.

PROPERTY SETTING DEFAULT REVISED

Cluster Features HA – Enabled

DRS – Enabled

vSphere HA Host Monitoring Status Enabled –

Admission Control Enabled –

Admission Control Policy Host failures the cluster tolerates=1

–

Virtual Machine Options > VM restart priority

Medium –

Virtual Machine Options > Host Isolation Response

Leave Powered On –

VM Monitoring Disabled –

Datastore Heartbeating Select any, taking into account my preferences (no datastore preferred)

–

vSphere Storage DRS Automation Level Fully automated (apply 1,2,3 priority recommendations)

–

DRS Groups Manager – –

Rules – –

Virtual Machine Options – –

Power Management Off –

Host Options Default (Disabled) –

Enhanced vMotion Capability

Disabled –

Swapfile Location Store in the same directory as the virtual machine

–

Table 2. vSphere Cluster Configuration



ESXi Servers

Each Virtual SAN ESXi server in the Virtual SAN cluster had the following configuration.

1x Intel 10GB 2P X520

1xLSI 9207-8i

1x 200GB IntelS3700 6Gbps SSD

Dell PowerEdge R7202x Intel Xeon E5-2690 @ 2.90Ghz

256GB RAM

4x 300GB 15K SA S 6Gbph

Figure 27. ESXi Host Components

PROPERTY SPECIFICATION

ESX server model 4 Dell R720BIOS v2.2.2 (1/16/2014)

ESXhostCPU [email protected]

ESX host RAM 256GB [16 x 16GB]

ESX version ESXi 5.5.0, 1746018

Network adapter Intel Ethernet 10G 2P X520Firmware version: 0x8000030dDriver version: ixgbe 3.7.13.7.14iov

Storage adapter LSI9207_8iFirmware version: 18.00Driverversion:MPT2SAS19.00.00.00.1vmwQueueDepth:600

Power management High Performance (set in BIOS)

Disks SSD: 1 Intel 200GB S3700 MLC SATA 6GbpsHDD: 4 Dell 300GB 15K RPM 2.5in 6Gbps SAS

Table 3. ESXi Host Configuration



Virtual SAN

The floating linked clones and replicas use Virtual SAN for storage. Each ESXi host had the same uniform configuration of a single disk group, consisting of one 200GB SSD and four 300GB 15K SAS disks. The SSD devices form the caching layer (fixed at a 70 percent read cache and 30 percent write buffer). Only the spinning magnetic disks contribute toward the usable storage capacity of the datastore. The four hosts yielded 4.8TB RAW.

4x Dell R7201 Disk Group Per Server

1x 200GB SATA 6Gbps SSD4x 300GB 15K RPM 6Gbps SAS

Virtual SAN Datastore

Figure 28. Virtual SAN Datastore Components

The virtual desktop replica did not need to be stored on a dedicated tier of flash storage because read I/O operations are served from the flash layer on Virtual SAN.

Virtual SAN Storage PolicyVirtual SAN can set availability, capacity, and performance policies per virtual machine if the virtual machines are deployed on the Virtual SAN datastore. The tests used the default storage policy settings, and VMware recommends using the defaults for View 5.3.2 virtual desktops. For Horizon 6 with View, specific storage-policy recommendations are based on pool type.

STORAGE CAPABILITY SETTING

Number of Failures to Tolerate 1

Number of Disk Stripes per Object 1

Flash Read Cache Reservation 0%

Object Space Reservation 0%

Table 4. Virtual SAN Storage Default Settings for View



Number of Failures to Tolerate (FTT) – This Virtual SAN storage protection policy is applied to each virtual machine. The FTT policy defines how many concurrent host, network, or disk failures can occur in the cluster and still ensure the availability of the object. The configuration contains at least FTT+1 copies of the virtual machine and a witness copy to ensure that the object’s data is available even when the number of tolerated failures occurs.

Figure 29. Default Virtual SAN Storage Policy – FTT=1

Object Space Reservation – By default, a virtual machine created on Virtual SAN is thin-provisioned, so it has no capability for object space reservation. It does not consume any capacity until data is written. You can change this setting between 0–100 percent of the virtual disk size. The virtual machine consumes this capacity of the Virtual SAN datastore when it is created.

The combination of the object space reservation percentage and the FTT settings applied to the virtual machines on the Virtual SAN datastore determines the datastore’s usable capacity.

Number of Disk Stripes per Object – This policy defines how many physical disks across each copy of a storage object are striped. The default value (recommended) of 1 was sufficient for our tested workloads.

Flash Read Cache Reservation – This is the amount of flash capacity reserved on the SSD as a read cache for the storage object. By default, all virtual machines share the read cache of an SSD equally.



Networking

A vNetwork distributed switch (dvSwitch) acted as a single vSwitch across all associated hosts in the data center. This setup allows virtual machines to maintain a consistent network configuration as they migrate across multiple hosts. The dvSwitch uses two 10GbE adapters per host.

VM VM vmk0

10GbE10GbE

Force10Switches

vmk1 vmk2

Desktop VLAN

ManagementVLAN

vMotion VLAN

Virtual SAN VLAN

vSphere Distributed Switch

Active LB based on physical NIC load

Figure 30. dvSwitch Configuration



Properties regarding security, traffic shaping, and NIC teaming can be defined on a port group. The following table shows the settings used with this design.

PROPERTY SETTING DEFAULT REVISED

General Port Binding Static –

Policies: Security Promiscuous mode Reject –

MAC address changes Accept Reject

Forged transmits Accept Reject

Policies: Traffic Shaping

Status Disabled –

Policies: Teaming and Failover

Load balancing Route based on the originating virtual port ID

Route based on physical NIC load

Failover detection Caution Link Status only –

Notify switches Yes –

Policies: Resource Allocation

Network I/O Control Disabled Enabled

Advanced MaximumMTU 1500 –

Table 5. Port Group Properties – dvSwitch v5 .5



Network I/O control was enabled for the distributed switch. The following settings and share values were applied on the resource allocation.

NETWORK RESOURCE POOL

HOST LIMIT (MBPS) PNIC SHARES SHARES

FT traffic Unlimited Normal 50

iSCSI Unlimited Normal 50

vMotion 8000Mbit/s Low 25

Management Unlimited Normal 50

NFS Unlimited Normal 50

Virtual machines Unlimited High 100

Virtual SAN Traffic Unlimited Normal 50

Table 6. Resource Allocations for Network Resources in dvSwitch

Horizon with View

The Horizon with View installation included the following core systems.

•Oneconnectionserver(N+1isrecommendedforproduction)

•OnevCenterServerwiththefollowingroles:

- vCenter

- vCenter single sign-on (SSO)

- vCenter Inventory Service

•ViewComposerrunningonvCenterServer

Note: Security servers were not used during this testing.

View Global PoliciesThe following global policies were in place for all system testing.

NETWORK RESOURCE POOL HOST LIMIT (MBPS)

USBaccess Allow

Multimedia redirection (MMR) Allow

Remote Mode Allow

PCoIP hardware acceleration Allow – medium priority

Table 7. View Global Policies



VMware View Manager Global Settings The following VMware View Manager™ global policies were used.

ATTRIBUTE SPECIFICATION

Session timeout 600 (10 hours)

View Administrator Session Timeout 120 minutes

AutoUpdate Enabled

Display pre-login message No

Display warning before logout Yes

Reauthenticate secure tunnel connections after network interruption

No

Enable IPSec for security server pairing Yes

Message security mode Enabled

Disable Single Sign-On for Local Mode Operations No

Table 8. View Manager Global Settings



vCenter Server SettingsView Connection Server uses vCenter Server to provision and manage View desktops. vCenter Server is configured in View Manager.


Description View vCenter Server

Connect using SSL Yes

vCenter Port 443

View Composer Port 18443

Enable View Composer Yes

Advanced Settings:Maximum Concurrent vCenter Provisioning OperationsMaximum Concurrent Power OperationsMaximum Concurrent View Composer Maintenance OperationsMaximum Concurrent View Composer Provisioning Operations

20501212

Storage Settings:Enable View Storage AcceleratorDefault Host Cache Size

√2048MB

Table 9. View Manager – vCenter Server Configuration



View Manager Pool SettingsThe following View Manager pool settings were used


Pool Type Automated Pool

UserAssignment Floating

Pool Definition – vCenter Server Linked Clones

Pool ID Desktops

Display Name Desktops

View folder /

Remote Desktop Power Policy Take no power action

Auto Logoff Time Never

UserResetAllowed False

Multi-Session Allowed False

Delete on logoff Never

Display Protocol PCoIP

Allow Protocol Override False

Maximum Number of Monitors 1

Max resolution 1920x1200

HTML Access Not selected

FlashQualityLevel Do not control

Flash Throttling Level Disabled

Enable Provisioning Enabled

Stop Provisioning on error Enabled

Provision all desktops up-front Enabled




Disposable File Redirection Do not redirect

Select separate Datastores for replica and OS Not selected

Datastores – Storage Overcommit Conservative

UseViewStorageAccelerator Selected

Reclaim VM disk space* N/A

Disk Types OS disks

Regenerate Storage Accelerator after 7 days

Reclaim VM Disk Space N/A

UseQuickprep Enabled

*Virtual SAN does not support the space-efficient (SE) sparse disk format.Table 10. View Manager – Test Pool Configuration



Test MethodologyThis reference architecture used Login VSI and View Planner to test performance and operational latency.

Login VSI 4.1 Workload Testing

Login Virtual Session Indexer (Login VSI) is the industry-standard benchmarking tool for measuring the performance and scalability of centralized desktop environments. Login VSI gradually increases the number of simulated users until saturation. When the system is saturated, the response time of the applications increases significantly. This latency indicates that the system is almost overloaded. Nearly overloading a system makes it possible to find out what its true maximum user capacity is.

After a test is performed, the response times can be analyzed to calculate the maximum active session per desktop capacity. This metric is called VSImax. When the system is approaching its saturation point, response times rise. By reviewing the average response time, you can see that the response time escalates at the saturation point.

VSI Launcher(s)

• Windows Desktop• Running View Agent• Microsoft Office and Login VSI Application Set• User Simulation Scripts

• Windows OS machine• Running Horizon Client

• Windows Domain Controller• Group Policy Objects for Login VSI• AD Users and Groups for View Pools

• Windows Server• Login VSI Binaries• Logfiles• Configuration

Viewdesktops

Active Directory Domain

(Optional)

VSI Share

Figure 31. Login VSI System Components

The simulated desktop workload is scripted in a 48-minute loop when a simulated Login VSI user is logged on performing generic office worker activities. After the loop is finished, it restarts. Within each loop, the response times of 12 operations are measured in a regular interval: 12 times within each loop. The response times of these operations are used to determine VSImax.



The operations from which the response times are measured are listed in the following table.

ID ACTION DESCRIPTION RELATED RESOURCES

WSLD Start Microsoft Word and load a random document

Word Start/Load a local random document file from content pool

CPU,RAM,and I/O

NSLD Start VSI-Notepad and load a document

VSI-Notepad Start/Load a local random text file from content pool

CPUandI/O

WFO Press file open in VSI-Notepad VSI-Notepad file open [Ctrl+O] CPU,RAM,and I/O

NFP Press print open in VSI-Notepad VSI-Notepad print open [Ctrl+P] CPU

ZHC Compress files with high compression

Compress a local random .pst file from content pool (5MB)

CPU

ZNC Compress files with no compression Compress a local random .pst file from content pool (5MB)

I/O

Table 11. Login VSI Operations

Login VSI has built-in workloads, so you can immediately start testing. Below are brief descriptions of the medium and heavy built-in workloads.

Medium WorkloadThe Login VSI default workload is medium. It emulates a medium knowledge worker using Microsoft Office, InternetExplorer,PDFfiles,Java,andFreeMind.ThemediumworkloadisdesignedtorunontwovCPUsperdesktop virtual machine. It has these characteristics:

•Afterasessionstarts,theworkloadlooprepeatsevery48minutes.

•Theloopisdividedinfoursegments.EachconsecutiveLoginVSIuserloginstartsatadifferentsegmenttoensure that all elements in the workload are equally used throughout the test.

•Duringeachloop,theresponsetimeismeasuredevery3–4minutes.

•Fiveapplicationsareopenedsimultaneously.

•Thekeyboardtyperateis160msforeachcharacter.

•Approximately2minutesofidletimeisincludedtosimulatereal-worldusers.

Each loop opens and uses:

•Outlooktobrowsemessages

•InternetExplorer,browsingdifferentWebpages;aYouTubestylevideo(480pmovietrailer)isopenedthreetimes in every loop

•Word,oneinstancetomeasureresponsetime,oneinstancetoreviewandeditadocument

•DoroPDFPrinterandAcrobatReader,printingandreviewingPDFfiles

•Excel,openingalargerandomizedsheet

•PowerPointtoreviewandeditapresentation

•FreeMind,aJava-basedmind-mappingapplication



Heavy WorkloadThe heavy workload requires the Login VSI PRO Content library, which includes 720p and 1080p videos.

In addition to what is performed in the medium workload, the heavy workload includes the following:

•BeginsbyopeningfourinstancesofInternetExplorer,whichremainopenthroughouttheworkloadloop.

•BeginsbyopeningtwoinstancesofAdobeReader,whichremainopenthroughouttheworkloadloop.

•MorePDFprinteractions.

•A720pvideoanda1080pvideoarewatched.

•IncreasesthetimeforplayingaFlashgame.

•Idletimeisreducedto2minutes.

View Planner 3.0 Workload Testing

VMware View Planner is a Virtual Desktop Infrastructure (VDI) workload generator that automates and measures a typical office user’s activity, such as using Microsoft Office applications, browsing the Web, and reading a PDF file. Each View Planner operation runs in an iterative fashion, and each iteration is a randomly sequenced workload. The results of a run consist of operational latencies collected for the applications and operations for all iterations.

VirtualAppliance

Physical Servers

Storage

Virtual Desktops

Manage

ClientManagement

DesktopManagement

Remote Display Protocol

vCenterView

WebInterface

ViewPlanner

Physical Servers

Storage

Virtual Client VMs

Harness

Figure 32. View Planner Components



ThestandardizedViewPlannerworkloadconsistsof9applicationsperformingatotalof44useroperations.These user operations are separated into three groups: interactive operations (Group A), I/O operations (Group B), and background load operations (Group C).

TheoperationsinGroupAareusedtodeterminequalityofservice(QoS).QoS,determinedseparatelyforGroupAuseroperationsandGroupBuseroperations,isthe95thpercentilelatencyofalltheoperationsinagroup. The default thresholds are 1.0 seconds for Group A and 6.0 seconds for Group B.

The operations in Group C are used to generate additional load.

GROUP A GROUP B GROUP C

AdobeReader: Browse AdobeReader: Open 7zip: Compress

AdobeReader: Close Excel_Sort:Open Outlook: Restore

AdobeReader: Maximize Excel_Sort:Save PowerPoint: SaveAs

AdobeReader: Minimize Firefox: Open Video: Play

Excel_Sort:Close IE_ApacheDoc:Open

Excel_Sort:Compute IE_WebAlbum:Open

Excel_Sort:Entry Outlook: Attachment-Save

Excel_Sort:Maximize Outlook: Open

Excel_Sort:Minimize PowerPoint: Open

Firefox: Close Video: Open

IE_ApacheDoc:Browse Word: Open

IE_ApacheDoc:Close Word: Save

IE_WebAlbum:Browse

IE_WebAlbum:Close

Outlook:Close

Outlook: Read

PowerPoint: AppendSlides

PowerPoint: Close

PowerPoint: Maximize



GROUP A GROUP B GROUP C

PowerPoint: Minimize

PowerPoint: ModifySlides

PowerPoint: RunSlideShow

Video: Close

Word: Close

Word: Maximize

Word: Minimize

Word: Modify

Table 12. View Planner Operations

For our testing, View Planner performed a total of five iterations:

•Rampup(firstiteration)

•Steadystate(second,third,fourthiterations)

•Rampdown(fifthiteration)

During each iteration, View Planner reports the latencies for each operation performed within each virtual machine.



Virtual Machine Test Image Build

Two different virtual machine images were used to provision desktop sessions in the View environment, one for View Planner testing and one for Login VSI workload testing. Both conformed to testing tool standards and were optimized in accordance with the View Optimization Guide for Windows 7 and Windows 8. The VMware OS Optimization Tool was used to make the changes.

ATTRIBUTE VIEW PLANNER LOGIN VSI

Desktop OS Windows 7 Enterprise SP1 (32-bit) Windows 7 Enterprise SP1 (32-bit)

Hardware VMwareVirtualHardwareversion9 VMwareVirtualHardwareversion9

CPU 1 2

Memory 1536MB 1536MB

Memory reserved 0MB 0MB

Video RAM 35MB 35MB

3D graphics Off Off

NICs 1 1

Virtual network adapter 1 VMXNet3 Adapter VMXNet3 Adapter

Virtual SCSI controller 0 Paravirtual Paravirtual

Virtual disk – VMDK 1 24GB 24GB

Virtual disk – VMDK 2 – 1GB

Virtual floppy drive 1 Removed Removed

Virtual CD/DVD drive 1 Removed Removed

Applications Adobe Acrobat 10.1.4Firefox 7.01Internet Explorer 10MS Office 2010

Adobe Acrobat 11Adobe Flash Player 11Doro PDF 1.82FreeMindInternet Explorer 10MS Office 2010

VMware Tools™ 9.4.5build-1734305 9.4.5build-1734305

VMware View Agent 5.3.2 5.3.2

Table 13. Virtual Machine Test Images

The Windows 7 Golden Image was modified in line with the View Planner 3 requirements. See Sections A, B, and C of the View Planner Installation and User’s Guide.

http://www.vmware.com/files/pdf/VMware-View-OptimizationGuideWindows7-EN.pdf

https://my.vmware.com/group/vmware/details%3FdownloadGroup%3DVIEW-PLAN-300%26productId%3D320



System SizingThis reference architecture used the following sizing specifications.

Hosts

Aspartofthesizingcalculations,itisimportanttofactorintheCPUandmemoryoverheadofVirtualSAN.VirtualSANisdesignedtointroducenomorethan10percentofCPUoverheadperhost.

DESKTOP PERFORMANCE METRIC RECORDED VALUE

AveragenumberofCPUsperphysicaldesktopsystem 1

AverageCPUutilizationperphysicaldesktopsystem 350MHz

vCPUoverhead 10%

Table 14. Host Sizing – Desktop CPU Requirements


NumberofCPUs(sockets)perhost 2

NumberofcoresperCPU 8

GHzperCPUcore 2.9GHz

TotalCPUGHzperCPU 23.2GHz

TotalCPUGHzperhost 46.4GHz

Virtual SAN overhead 10%

AvailableCPUGHzperhost 41.76GHz

Desktops per host 108

Table 15. Host Sizing – CPU

ItisrecommendedtoallowsomeheadroomforCPUspikes,hostfailures,andmaintenancewithinyourvSphereclusters.



The host memory requirement depends on various factors, including memory allocated per virtual desktop, virtualdesktopgraphicsrequirements,vCPUmemoryoverhead,VirtualSANoverheadof10percent,andvSphere transparent page sharing (TPS).


Virtual SAN memory usage 10%

Total amount of RAM per virtual machine 1536MB

VMCPUNo’s 1

Anticipated TPS saving 30%

VM Memory Reservation 0%

VM Resolution 1x[1920x1600]

VM Video Memory 8.79MB

3D -

VM Memory Overhead 35.27MB

Total amount of RAM per host 207GB TPS

Table 16. Host Sizing – Memory

Virtual SAN

Virtual SAN introduces some new constructs that have sizing requirements.

Disk GroupsA disk group is a container for magnetic disks and an SSD that acts as a read cache and write buffer. Each disk group must have one SSD and at least one magnetic disk, with a maximum of seven. Each ESXi host can have up to five disk groups.

The type of magnetic disk depends on the level of performance and capacity required. For linked clones, it is recommended to have at least three magnetic 10K or 15K disks per disk group.

When an individual magnetic disk is 80 percent utilized, Virtual SAN rebalances the components to other magnetic disks. This action incurs a performance overhead on the cluster.

The recommended sizing for the SSD is 10 percent of the total consumed storage capacity (excluding FTT). For example, a pool of 400 linked-clone desktops is expected to use 5GB of disk space per virtual desktop:

10% (400 x 5) = 200GB

With a Virtual SAN cluster of four hosts, the minimum recommended SSD size is 50GB per host.

Only magnetic disks count toward cluster capacity. Cluster capacity equals:

Num_hosts x Num_disk_groups x Num_disks_per_group x disk_size



For our test configuration, the capacity is:

4 x 1 x 4 x 300 = 4800GB = 4.8TB

Objects and ComponentsVirtual SAN objects include the virtual machine home (namespace), virtual machine swap files, virtual machine disk format (VMDK) files, and snapshots. The namespace includes VSA files, log files, and virtual machine configuration files. The default number of objects per virtual desktop type is listed in the following table.

USER ASSIGNMENT

VIRTUAL MACHINE TYPE

DISPOSABLE DISK

OBJECTS NUMBER OBJECTS PER DESKTOP

NUMBER COMPONENTS PER DESKTOP

Floating Linked clone

N Namespace, VMDK, snapshot, internal disk, swap

5 9replicaplus15 per virtual machine

Floating Linked clone

Y Namespace, VMDK, snapshot, internal, disposable, swap


Dedicated Linked clone

Y Namespace, VMDK, snapshot, internal disk, disposable disk, persistent disk, swap


Floating Full clone – Namespace, VMDK, swap

3 9

Dedicated Full clone – Namespace, VMDK, swap

3 9

Table 17. Default Number of Objects per View Desktop Type

Note: Adjusting the default Space Policy-Based Management policies alters the component counts listed here.

For our test configuration, 400 floating, linked-clone virtual machines with no disposable disk comes to objects per desktop. The total number of objects for our pool of 400 virtual machines is:

400 x 5 = 2000

The number of objects per virtual machine, in addition to their performance and availability requirements, dictates the number of components that will be created. Virtual SAN supports a maximum of 3,000 components per host.

The default storage policies for Virtual SAN and Horizon with View are as follows. VMware recommends using the default policies.

•FTT=1

•Numberofdiskstripesperobject=1



FTT defines the number of hosts, disk, or network failures that a storage object can tolerate. For n failures tolerated, n+1 copies of the object are created, and 2n+1 host contributing storage is required.

FTT has the greatest impact on capacity in a Virtual SAN cluster. Based on the availability requirements of a virtual machine, the setting defined in a virtual machine storage policy can lead to the consumption of several multiples of the virtual desktop consumed space.

The number of disk stripes per object is the number of HDDs across which each replica of a storage object is distributed.

You can use the following formula to calculate the number of components per desktop. It accounts for the replicas and witnesses created based on the FTT setting. The resulting number of components is split across all the hosts in the cluster.

Number of components = Objects x [FTT x 2 + 1]

For our configuration this is 2000 x [(1 x 2) +1] = 6000 components.

Inaddition,threeobjects(namespace,swap,andVMDK)arecreatedforthereplicavirtualdisk.WithFTT=1,thatis9additionalcomponents,so6,009intotal.

Management Blocks

SERVER ROLE VCPU RAM (GB) STORAGE (GB) OS

Domain Controller 2 6 40 Server 2008 64-bit R2

SQLServer 2 8 140 Server 2008 64-bit R2

vCenter Server 4 10 50 Server 2008 64-bit R2

View Connection Server

4 10 60 Server 2008 64-bit R2

vCenter Operations Manager Analytics Server

8 21 212 SLES 11–64-bit

ManagerUIServer 8 13 132 SLES 11–64-bit

Table 18. Management Block Sizing



The following table shows the peak resource usage by management block components throughout workload, operations, and resiliency testing.

RESOURCE MAX CPU (MHZ)

MAX MEM USAGE (MB)

MAX NETWORK RX/TX (KBPS)

MAX DISK I/O (KBPS)

MAX DISK I/O OPERATIONS [PS]

View Manager 3826MHz 6982MB 206/109KBps 8861KBps R437 / W32

VMware vCenter Operations [Analytics virtual machine]

6439MHz 20468MB 216/1995KBps 26252KBps R890/W63

VMware vCenter Operations [UIvirtualmachine]

2193MHz 13191MB 1992/377KBps 5852KBps R358 / W80 IOPS

VMware vCenter

7421MHz 10314MB 9089/8964KBps 13456KBps R461 / W645 IOPS

SQLServer 963MHz 6148MB 467/289KBps 3790KBps R33/W89IOPS

Domain Controller

1010MHz 3197MB 40 / 82KBps 3558KBps R9/W50IOPS

Table 19. Management Peak Resource Usage



Bill of MaterialsThe following table summarizes the bill of materials for this reference architecture.

AREA COMPONENT QUANTITY

Host hardware Dell PowerEdge R720 4

[email protected] 2

256GB RAM

LSI9207_8i 1

Intel Ethernet 10G 2P X520 1

1GB SD Card 1

Intel 200GB S3700 MLC SATA 6GBps SSD 1

Dell 300GB 15K RPM 2.5in 6Gbps SAS 4

Network switch Dell Force10 S4810 10/40GbE 2

Software VMware ESXi 5.5, 1746018 7

VMware vCenter Server 5.5.0, 1623101 1

View 5.3.2 1

VMware vCenter Operations for View 1.5.1, 1286478 1

Microsoft Windows 2008 R2 4

MicrosoftSQLServer2008R2 1

Table 20. Bill of Materials



ConclusionVMware Virtual SAN is a low-cost, high-performance storage platform that is rapidly deployed, easy to manage, and fully integrated into the industry-leading VMware vSphere cloud suite.

It scales as your VDI user base does, keeping CapEx costs down and eliminating the need for the large upfront investment that traditional storage arrays often require.

Extensive workload, operations, and resiliency testing shows that Horizon with View on Virtual SAN delivers high levels of performance, a great end-user experience, and solid system resiliency, all at a low price point.



About the AuthorsMattCoppingerisagroupproductmanagerwiththeVMwareEnd-UserComputing(EUC)TechnicalMarketingGroup, with a focus on best practice technical architecture and customer implementations. Previously, Matt ledtheGlobalEUCConsultingPracticeatVMware,participatedinlarge-scaleEUCdeploymentsinEMEA,andworked as an engineer on the original View product.

DonalGearyworksasaReferenceArchitectEngineerwiththeVMwareEnd-UserComputingTechnicalMarketing Group, with a focus on performance testing. Previously he worked as Technical Lead for VMware IT’s EMEA and APAC regions, and as Escalation Engineer for the VMware Global Support Services Organization. He holds a BSc. in Applied Computing and is a certified Cloud Architect Expert (EMCCAe) and Storage Networking Specialist (EMCDCA).

Wade Holmes, VCDX #15, CISSP, CCSK, is a senior technical marketing architect at VMware, currently focusing on software-defined storage reference architectures. He has more than 17 years of industry experience in the design and implementation of complex computing environments of all scopes and sizes. He has presented at many industry conferences and is a co-author of VMware vCloud Architecture Toolkit. He holds a bachelor’s degree in information technology and a master’s degree in information assurance.

Follow Wade’s blogs:

•http://blogs.vmware.com/vsphere/storage

• http://www.vwade.com

Follow Wade on Twitter: @wholmes

http://blogs.vmware.com/vsphere/storage

http://www.vwade.com

VMware, Inc. 3401 Hillview Avenue Palo Alto CA 94304 USA Tel 877-486-9273 Fax 650-427-5001 www.vmware.comCopyright © 2014 VMware, Inc. All rights reserved. This product is protected by U.S. and international copyright and intellectual property laws. VMware products are covered by one or more patents listed athttp://www.vmware.com/go/patents. VMware is a registered trademark or trademark of VMware, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. Item No: VMW-TWP-VIRTSANREFERARCH-USLET-20140722-WEB


ReferencesView Documentation

View Technical Resources

View Optimization Guide for Windows 7 and Windows 8

View Planner 3 Resources

View Storage Accelerator

VMware vCenter Database Performance Improvements + Best Practice for Large-Scale Environments

Dell PowerEdge R720

Dell Force10 S4810 High-performance 10/40GbE top-of-rack switch

Intel Ethernet Server Adapter X520-T2

LSI SAS 9207-8i

Virtual SAN Sizing Guide for VDI and other technical resources

Virtual SAN Hardware Quick Reference Guide

Virtual SAN Design and Sizing Guide

VDI Performance Benchmarking on VMware Virtual SAN 5.5

https://www.vmware.com/support/pubs/view_pubs.html

http://www.vmware.com/products/desktop_virtualization/view/technical-resources.html

http://www.vmware.com/resources/techresources/10157

https://my.vmware.com/group/vmware/details?downloadGroup=VIEW-PLAN-300&productId=320

http://www.vmware.com/resources/techresources/10334

http://www.vmware.com/files/pdf/techpaper/VMware-vCenter-DBPerfBestPractices.pdf

http://www.dell.com/downloads/global/products/pedge/dell-poweredge-r720-spec-sheet.pdf

http://i.dell.com/sites/doccontent/shared-content/data-sheets/en/Documents/Dell_Force10_S4810_Spec_sheet.pdf

http://www.dell.com/learn/us/en/19/shared-content~data-sheets~en/documents~nic-intel-ethernet-x520-da2-pb.pdf

http://www.lsi.com/products/host-bus-adapters/pages/lsi-sas-9207-8i.aspx

http://www.vmware.com/products/virtual-san/resources.html

http://partnerweb.vmware.com/programs/vsan/Virtual%20SAN%20Hardware%20Quick%20Start%20Guide.pdf

https://www.vmware.com/files/pdf/products/vsan/VSAN_Design_and_Sizing_Guide.pdf

http://blogs.vmware.com/performance/tag/view-planner