xendesktop planning guide - dabcc.com - planning guide - microsoft...xendesktop planning guide ......

44
Worldwide Consulting Solutions | WHITE PAPER | Citrix XenDesktop www.citrix.com XenDesktop Planning Guide Integration with Microsoft Hyper-V 2008 R2

Upload: ngokien

Post on 21-May-2018

219 views

Category:

Documents


1 download

TRANSCRIPT

Worldwide Consulting Solutions | WHITE PAPER | Citrix XenDesktop

www.citrix.com

XenDesktop Planning Guide

Integration with Microsoft Hyper-V 2008 R2

Page 2

Table of Contents

Introduction ........................................................................................................................................................ 5

Reference Architecture ...................................................................................................................................... 6

Conceptual Architecture ............................................................................................................................... 6

Design Decision Summary ........................................................................................................................... 7

Architectural Guidelines ................................................................................................................................. 10

Hardware ....................................................................................................................................................... 10

1. Hardware Compatibility ................................................................................................................. 10

2. Hyper-V Edition .............................................................................................................................. 10

3. Processor Specification................................................................................................................... 11

4. Memory Capacity ............................................................................................................................ 11

5. Storage Capacity .............................................................................................................................. 12

6. Direct-attached Storage .................................................................................................................. 12

7. Shared Storage Connectivity .......................................................................................................... 13

8. Network ............................................................................................................................................ 14

9. Scale Up/Out ................................................................................................................................... 15

10. Hyper-V Host Scalability ........................................................................................................... 15

11. Dynamic Memory ....................................................................................................................... 16

12. Component Redundancy ........................................................................................................... 17

Networking ................................................................................................................................................... 17

1. Physical Network Connections ..................................................................................................... 17

2. NIC Teaming ................................................................................................................................... 18

3. Virtual Adapter ................................................................................................................................ 18

4. Provisioning Services ...................................................................................................................... 19

5. IP Addressing................................................................................................................................... 19

6. VLANs .............................................................................................................................................. 19

Page 3

7. Security .............................................................................................................................................. 20

Storage ........................................................................................................................................................... 20

1. Direct-attached Storage/Storage-attached network ................................................................... 20

2. Database Storage ............................................................................................................................. 21

3. Tiered Storage .................................................................................................................................. 21

4. Performance ..................................................................................................................................... 21

5. Thin Provisioning ............................................................................................................................ 23

6. Data De-Duplication ...................................................................................................................... 24

Hyper-V Failover Cluster ............................................................................................................................ 24

1. Virtual Machines to Cluster ........................................................................................................... 24

2. Number of Failover Clusters ......................................................................................................... 25

3. Nodes per Failover Cluster ............................................................................................................ 26

4. Cluster Shared Volumes ................................................................................................................. 26

System Center 2012 Virtual Machine Manager ........................................................................................ 27

1. VMM Architecture .......................................................................................................................... 27

1. VMM Design and Scalability ......................................................................................................... 28

Operating System Delivery ............................................................................................................................. 30

1. Templates ......................................................................................................................................... 30

2. Machine Creation Services ............................................................................................................. 30

3. Provisioning Services ...................................................................................................................... 31

4. Desktop Customization.................................................................................................................. 33

Management ..................................................................................................................................................... 34

Monitoring .................................................................................................................................................... 34

1. Management Tool ........................................................................................................................... 34

2. Management Tool Access .............................................................................................................. 34

3. Hyper-V Hardware ......................................................................................................................... 34

Page 4

4. Hyper-V Performance .................................................................................................................... 34

5. XenDesktop Infrastructure............................................................................................................ 35

Backup and Recovery .................................................................................................................................. 35

1. Backup Software .............................................................................................................................. 35

2. Backup Type .................................................................................................................................... 35

3. Components to Backup .................................................................................................................. 35

High Availability ........................................................................................................................................... 36

1. VMM server ..................................................................................................................................... 36

2. VMM Library ................................................................................................................................... 36

3. SQL Server ....................................................................................................................................... 37

4. XenDesktop Infrastructure VMs .................................................................................................. 37

5. Virtual Desktops .............................................................................................................................. 37

Appendix ........................................................................................................................................................... 37

Appendix A: Optimizations........................................................................................................................ 37

Appendix B: High-Level Implementation Steps ..................................................................................... 39

Product Versions.............................................................................................................................................. 44

Revision History ............................................................................................................................................... 44

Page 5

Introduction

This document provides design guidance for Citrix XenDesktop 5.x deployments that leverage

Microsoft Hyper-V 2008 R2. It should not be considered as a replacement for other Hyper-V or

XenDesktop design guidance, but rather an addendum that will assist in design decisions specifically

related to integrating XenDesktop with Microsoft Hyper-V.

It is assumed that the reader is already familiar with the architecture of XenDesktop. For an

overview of the XenDesktop architecture, please refer to the Citrix Knowledgebase Article

CTX133162 – Modular Reference Architecture.

The document has been divided into four sections –

1. Reference Architecture. Provides a high-level conceptual architecture and default

recommendations for key design decisions that are discussed in detail within the

Architectural Guidelines section of the document.

2. Architectural Guidelines. Provides guidance on the core components of the Hyper-V

architecture including host hardware, networking configuration, storage, clustering and the

VMM server.

3. Operating System Delivery. Examines the key Hyper-V design decisions relating to

operating system delivery; options include – Virtual Hard Disks, Machine Creation Services

and Provisioning Services.

4. Management. Discusses the key Hyper-V design decisions that need to be considered so

that the XenDesktop environment is stable, highly available and continues to provide users

with a high-level of performance.

After reading this document, the reader will be able to effectively design a XenDesktop deployment

on Microsoft’s hypervisor and have a high-level understanding of the key implementation steps. For

further planning guides, please refer to the XenDesktop Design Handbook.

Page 6

Reference Architecture

Conceptual Architecture

The following diagram depicts a typical Hyper-V architecture used to support XenDesktop where

separate Hyper-V clusters have been created to host the infrastructure servers and dedicated virtual

desktops. The Hyper-V hosts supporting the pooled desktops and XenApp servers have not been

clustered as they are user persistent. If a host fails, identical virtual desktops will still be available on

another host. This allows costs to be reduced by using cheap local storage for non-persistent

desktops.

Page 7

Design Decision Summary

The following table provides default recommendations for the key Hyper-V design decisions. These

recommendations are intended to provide a starting point that is valid in the majority of situations.

Additional information on each design decision is available within the Architectural Guidelines

section of this document.

Decision Default Recommendation Justification

HARDWARE

Hyper-V Edition for

virtual desktops

Microsoft Hyper-V Server 2008 R2

Microsoft Hyper-V Server 2008 R2 is a

good choice for virtual desktops because

it has a reduced attack surface and

requires less maintenance than Windows

Server 2008 R2.

Hyper-V Edition for

XenApp servers and

infrastructure servers

Microsoft Windows Server 2008 R2

Datacenter

Microsoft Windows Server 2008 R2

Datacenter is a good choice for the

XenApp servers and infrastructure

servers because it includes unlimited

virtual server image use rights.

Processor

Specification

1. 64-bit Intel VT or AMD-V

processors

2. Hardware Data Execution

Protection (DEP)

3. Second Level Address

Translation (SLAT) capable

processors

Processors that support this

specification will support Hyper-V and

offer an improved level of scalability.

RAID Configuration RAID-1 / RAID-10 Hyper-V hosts supporting the write

intensive Provisioning Services write-

cache (80% write) and Machine Creation

Services difference disks (50% write)

should utilize either RAID-1 or RAID-

10 to help minimize write penalties.

Network Card

Specification

1. TCP chimney offload

2. Virtual Machine Queue (VMQ)

3. Jumbo Frames

4. 1Gbps or greater

Network cards that meet this

specification will help to ensure that the

network does not become a bottleneck

and also improve host scalability.

Scaling Up/Out Scale Up Scaling up typically results in a less

expensive solution due to a reduction in

power, networking, management,

support and hosting costs.

Dynamic Memory Enabled Provides additional VM density per host

when memory is the limiting factor.

Redundancy Redundant storage and network

connections as well as power

supplies, fans and disks.

Prevents a single component failure

from impacting a large number of users.

Page 8

NETWORKING

Physical Networks 4 x Network

2 x NICs (teamed): Live Migration Network, Cluster Shared Volume Network, Virtual Machine Network, Provisioning Network (teamed)

1 x NIC: Management Network

1 x NIC: Cluster Private Network

Assumed that HBA connections used for shared storage.

Virtual Network

Adapter Type

Legacy The legacy network connection will be

suitable for all but the most network

intensive environments.

LAN / VLANs 1. Desktop Network

2. Server Network

3. Provisioning Network

For reasons of security and

performance, the XenDesktop

infrastructure should be hosted on a

separate VLAN to the virtual desktops.

Separate physical networks are typically

created for management, cluster and

storage networks due to security and

bandwidth requirements.

STORAGE

Direct-attached

Storage

1. Host partition install

2. XenApp servers

3. Pooled desktops

The XenApp servers and pooled

desktops do not typically contain user

specific data or applications which

require the use of a failover cluster or

live migration.

Storage-attached

Network

1. Dedicated Desktops

2. Infrastructure servers

Components that require a high-level of availability should be stored on the SAN so that live migration and failover clustering can be used.

Virtual machines per

LUN

30 Limiting the number of virtual machines per LUN helps to ensure that the iSCSI queue does not become congested.

Thin Provisioning Disabled Microsoft does not recommend that

system volumes or page files are hosted

on thin provisioned LUNs.

De-Duplication Enabled Reduces storage requirements.

Monitoring should be implemented to

ensure that de-duplication does not

impact performance.

HYPER-V FAILOVER CLUSTERS

Virtual Machines to

Cluster

1. Infrastructure servers

2. Dedicated desktops

Typically require a high-level of

availability.

Page 9

Number of Clusters Minimum of 2 –

1. Server infrastructure

2. Dedicated virtual desktops

Ensure that dedicated virtual desktop

resources are isolated from

infrastructure resources for optimal

performance and availability.

Resilience per cluster N+1 Ensures that there is sufficient capacity within the cluster to accommodate a single host becoming unavailable or being taken offline for maintenance.

OPERATING SYSTEM DELIVERY

Operating System

Delivery Technology

Provisioning Services Provisioning Services is recommended

as Hyper-V is not supported with

Intellicache. As such, Machine Creation

Services generates approximately 50%

more IOPS than Provisioning Services.

Personal vDisk Enabled for power users and

knowledge workers

These user groups typically require the

ability to customize their desktop

(beyond profile changes).

BACKUP & RECOVERY

Backup Type Backup from Hyper-V host Performing the backup from the Hyper-

V host is preferred because it captures

more data than performing the backup

from within the guest operating.

HIGH AVAILABILITY

VMM Server Hosted on Huper-V Failover Cluster Required so that XenDesktop can

control the power state of the virtual

desktops.

SQL High

Availability

Hosted on Hyper-V Failover Cluster Required so that support staff can

administer the Hyper-V environment.

VMM Library No High Availability The VMM Library is not essential for

the availability of existing virtual

desktops or XenApp servers.

Infrastructure Virtual

Machines

Hosted on Hyper-V Failover Cluster Ensures that a single host failure does

not affect the availability or

performance of the XenDesktop

environment.

Virtual Desktop

Hosts

Dedicated desktops hosted on

Microsoft Failover Cluster.

Dedicated desktops are unique, unlike

pooled desktops.

Page 10

Architectural Guidelines

The sections below include architectural guidance and scalability considerations that should be

addressed when planning a XenDesktop deployment hosted on Microsoft Hyper-V.

Hardware

The hardware selected for the Hyper-V hosts has a direct impact on the performance, scalability,

stability and resilience of the XenDesktop solution. As such, the following key design decisions

must be considered during the hardware selection process:

1. Hardware Compatibility

For reasons of stability, it is imperative that the Hyper-V hardware specification selected in the

design is compatible with Hyper-V. Microsoft maintains a list of servers and hardware

components capable of supporting Hyper-V through the Windows Server Catalog website.

Servers and hardware components with the “Certified for Windows Server 2008 R2” and

“Certified for Windows Server 2008” logo pass the Microsoft standards for supporting x64 and

virtualization technologies.

2. Hyper-V Edition

The design must determine which variation of Hyper-V will be used:

Hyper-V Server 2008 R2. Standalone version of Hyper-V that does not include a

Graphical User Interface (GUI). Administration is either performed remotely, or via the

command line. Hyper-V Server does not require a Windows 2008 R2 license, however

licenses are required for each guest virtual machine. For more information, please refer

to the Microsoft Hyper-V Server 2008 R2 website.

Windows Server 2008 R2. Hyper-V is also available as an installable role for Windows

Server 2008 R2 Standard, Enterprise and Datacenter editions, each of which includes

different levels of “virtual server image use rights” which range from one Microsoft

server operating system license to unlimited. For more information, please refer to the

Microsoft Windows Server 2008 R2 website.

The following table highlights the key differences between Hyper-V Server 2008 R2 and the

different editions of Windows Server 2008 R2 when it comes to supporting XenDesktop:

Capability Hyper-V Server 2008 R2

Windows Server 2008 R2

Standard Enterprise Datacenter

Virtual Server Image Use Rights 0 1 4 Unlimited

Sockets 8 4 8 64

Memory 1TB 32GB 1TB 1TB

Cluster Shared Volumes Yes No Yes Yes

Page 11

Hyper-V Server 2008 R2 is typically preferred for XenDesktop deployments because it has a

decreased attack service, improved stability, reduced maintenance and does not require a

Windows Server 2008 R2 license. Windows Server 2008 R2 Enterprise and Datacenter is

recommended when familiarity with the command-line is limited or there is third-party software

to be installed on the server that is not supported on the Hyper-V Server 2008 R2 version.

The virtual server image rights included with Windows Server 2008 R2 make it an excellent

choice for hosting XenApp servers.

Note: The Standard edition of Windows Server 2008 R2 is rarely used due to the limited memory support. In

addition, the Standard edition lacks Cluster Shared Volumes support which is a recommended feature for hosting

infrastructure servers and dedicated virtual desktops.

3. Processor Specification

The Hyper-V design must ensure that the processor specification selected is capable of

supporting Hyper-V:

64-bit Intel VT or AMD-V processors with a clock speed of at least 1.4 GHz, however

2GHz or faster is recommended.

Hardware Data Execution Protection (DEP), specifically Intel XD bit (execution disable

bit) or AMD NX bit (no execute bit), must be enabled and available. For more

information, please refer to Microsoft TechNet Article 912923 - How to determine that

hardware DEP is available and configured on your computer.

For improved performance and scalability, ensure that Second Level Address Translation

(SLAT) capable processors are selected. Intel calls this technology Extended Page Tables (EPT)

and AMD calls it Nested Page Tables (NPT). The technology provides hardware assistance for

memory virtualization, memory allocation, and address translation between the physical and

virtual memory. The reduction in processor and memory overhead improves scalability allowing

a single Hyper-V host to run more virtual machines.

Note: Live Migration allows administrators to move virtual machines between physical hosts. In order to use this feature, all physical host servers in a cluster must have processors from the same family.

4. Memory Capacity

The hardware selected for the Hyper-V hosts must have sufficient memory to support the

parent partition as well as the guest virtual machines that they support:

At a minimum, the parent partition requires 512MB of RAM, however 2GB or greater is

recommended for VDI environments.

Virtual desktops typically require between 768MB and 4GB of RAM depending on

workload and operating system used. For more information, please refer to the Hosted

Page 12

VDI Scalability section in the Citrix Knowledgebase Article CTX133162 – XenDesktop

Module Reference Architecture.

Each Citrix XenApp user typically requires between 200MB and 1GB of RAM

depending on their workload. For more information, please refer to the Hosted Shared

Scalability section in the Citrix Knowledgebase Article CTX133162 – XenDesktop

Module Reference Architecture.

Note: With the exception of Microsoft Windows Server 2008 R2 Standard edition which is limited to 32GB, Hyper-V can support up to a total of 1TB of physical memory.

5. Storage Capacity

Sufficient storage must be specified for the parent partition and any guest VMs that are to be

hosted on local storage (typically pooled desktops and XenApp servers):

At a minimum, Hyper-V Server requires 8GB of storage, however, 20GB or greater is

recommended. For more information, please refer to the Microsoft Website – Hyper-V

Server 2008 R2 System Requirements.

Microsoft Windows Server 2008 R2, requires a minimum of 32GB of storage, however,

40GB or greater is recommended. For more information, please refer to the Microsoft

TechNet Article – Installing Windows Server 2008 R2.

Virtual desktops typically require between 10 and 20GB of storage depending on

application set and operating system used. Provisioning Services helps to reduce virtual

desktop storage requirements to between 1 and 3GB.

Virtual XenApp servers typically require between 40 and 60GB of storage depending on

the application set. Provisioning Services typically helps to reduce the XenApp storage

requirements to between 10 and 15GB.

The first time a virtual machine is started Hyper-V creates a .BIN file that is equal in size

to the RAM configured for the virtual machine. This file is used to store the contents of

the RAM when the virtual machine is moved into a saved state. The file is created

automatically in the same location where the virtual machine resides and cannot be

disabled.

6. Direct-attached Storage

The Direct-attached Storage (DAS) specification selected during the design has a direct impact

on both performance and availability of the XenDesktop environment, particularly when it is

used to host guest VMs (typically pooled desktops and XenApp servers):

For optimal XenDesktop performance, it is imperative that the DAS solution can handle the IO generated by the virtual desktops on the host:

Page 13

FlexCast Model Performance Category Steady State IOPS per User

Hosted VM Light 8

Normal 15

Heavy 50

Hosted Shared Light 2

Normal 4

Heavy 8

IOPS performance can be improved by adding disks to the RAID array and/or using high performance disks (e.g. SAS 15k / SSD). In addition it is highly recommended that the array controllers are specified with a battery backed cache and a minimum of 512MB of memory. For more information, please refer to the Citrix Knowledgebase Article CTX130632 – Storage Best Practices.

Using SSDs is particularly critical in environments with blade servers because most vendors only provide two slots for local storage per blade.

Note: Hyper-V supports the following Direct-attached Storage (DAS) technologies - Serial Advanced Technology Attachment (SATA), external Serial Advanced Technology Attachment (eSATA), Parallel Advanced Technology Attachment (PATA), Serial Attached SCSI (SAS), SCSI, USB, and Firewire.

While some RAID levels are optimized for writes (i.e. RAID 0, 1, 10) others such as

RAID 5 or 6 are not:

RAID Level Capacity Read Performance (Random)

Write Performance (Random)

RAID-0 100% Very Good Very Good (Write Penalty 0)

RAID-1 50% Very Good Good (Write Penalty 2)

RAID-5 Disk Size * (number of disks -1)

Very Good Bad (Write Penalty 4)

RAID-10 50% Very Good Good (Write Penalty 2)

Therefore, Hyper-V hosts supporting the write intensive Provisioning Services write-

cache (80% write) and Machine Creation Services difference disks (50% write) should

utilize either RAID-1 or RAID-10. RAID-0 is not recommended because it does not

offer redundancy.

7. Shared Storage Connectivity

Hyper-V hosts that support virtual machines on shared storage (typically dedicated desktops and

infrastructure servers) must have an appropriate storage connection. Hyper-V supports the use

of either Host Bus Adapters (HBA) or Ethernet connections to Storage Area Networks.

Note: Hyper-V supports the following Storage Area Network (SAN) technologies - Internet SCSI (iSCSI), Fibre Channel, and SAS technologies.

Page 14

Note: Microsoft does not support Network Attached Storage (SMB, CIFS or NFS protocols) with Hyper-V. For more information, please refer to Microsoft Knowledgebase Article KB2698995 – Microsoft Support Policy for Hyper-V Environments Utilizing Network Attached Storage.

8. Network

For reasons of performance and resiliency, it is recommended that the following

recommendations are taken into consideration when selecting the Hyper-V host network cards:

Sharing network cards between virtual desktops can lead to a network bottleneck. The

use of fast network adapters/switches (1Gbps or greater) will help prevent the network

from becoming a bottleneck.

If sufficient infrastructure exists, network throughput can be increased by separating

different types of network traffic across multiple physical NICs, for example

management, virtual machine, cluster heartbeat, live migration, provisioning, backup and

storage traffic can all be isolated from each other. Please refer to the Networking

section of this document for further details.

The implementation of NIC teaming provides increased resiliency by allowing multiple

network cards to function as a single entity. In the event of one NIC failing, traffic will

be automatically routed to another NIC in the team.

Note: Microsoft does not support NIC teaming with Hyper-V; however, third party solutions are

available. For more information, please refer to Microsoft Knowledgebase Article KB968703 -

Microsoft Support Policy for NIC Teaming with Hyper-V.

Virtual Machine Queue (VMQ) capable network cards should be selected to improve

network performance. VMQ delivers network traffic from an external virtual machine

directly to the virtual machines running on the host, thus reducing the overhead from

routing packets.

TCP chimney offload capable network adapters should be selected so that network stack

processing can be offloaded to the network adapter. For more information, please refer

to the Microsoft Knowledgebase Article KB951037 – Information about the TCP

Chimney Offload.

Jumbo Frames compatible network cards should be selected for dedicated iSCSI

connections to the SAN. Jumbo frames are Ethernet frames containing more than 1500

bytes of payload. Jumbo frames allow higher throughput to be achieved, reducing the

load on system bus memory, and reducing the CPU overhead. For more information,

please refer to the Microsoft Blog post – Hyper-V Network Optimizations.

Note: All NICs and switches between the Hyper-V hosts and the storage solution must support jumbo

frames.

Page 15

For details on the number of network cards required, please refer to the Physical Network

Connections recommendations within the Networking section of this document.

9. Scale Up/Out

There are a number of environmental factors that need to be considered when determining

whether the Hyper-V host hardware specification should be “scaled up” (reduced number of

high-performance hosts) or “scaled out” (increased number of less powerful hosts), including:

Data Center Capacity. The data center may have limited space, power and/or cooling

available. In this situation, consider scaling up.

Infrastructure and Maintenance Costs. When determining the overall cost of the

Hyper-V hosts, careful consideration should be taken when planning to include costs

such as rack space, support contracts and the network infrastructure required.

Hosting Costs. There may be hosting and/or maintenance costs based on the number

of physical servers used. If so, consider “scaling up” to reduce the long-term costs of

these overheads.

Redundancy. Spreading user load across additional less-powerful servers helps reduce

the number of users affected from hardware or software failure on a single host. If the

business is unable to accept the loss of a single high-specification server, consider

“scaling out”.

10. Hyper-V Host Scalability

During the design, it’s necessary to estimate the number of physical host servers that will be

required to support the XenDesktop implementation. Ideally, scalability testing should be

performed prior to the hardware being ordered, however this is not always possible. In these

situations, consider the following guidelines when determining single host scalability:

The parent partition typically utilizes between 8 and 12% of the processor resources

available leaving between 88 and 92% available for the child partitions.

At a minimum, the parent partition requires 512MB of RAM, however 2GB or greater is

recommended for VDI environments.

Microsoft supports up to 12 virtual processors per logical processor for Hyper-V hosts

supporting Windows 7. A maximum of 8 virtual processors per logical processor is

supported for all other operating systems.

Physical processors

Cores per processor

Threads per core (Hyper-Threading)

Max Virtual Processors – Win7

Max Virtual Processors – All Other OS

2 2 2 96 64

Page 16

2 4 2 192 128

2 6 2 288 192

2 8 2 384 256

2 10 2 480 320

4 2 2 192 128

4 4 2 384 256

4 6 2 512 (Maximum) 384

4 8 2 512 (Maximum) 512 (Maximum)

4 10 2 512 (Maximum) 512 (Maximum)

It is important to realize that these are maximum values. Testing from Cisco shows that a Cisco

UCS B230 M2 Blade Server (2 processor / 10 core / 2 threads) supports 145 Windows 7

desktops running the Login VSI medium workload (rather than the 480 maximum supported).

Processor was identified as the primary bottleneck. For more information, please refer to the

Cisco Whitepaper – Citrix XenDesktop on FlexPod with Microsoft Private Cloud.

11. Dynamic Memory

Citrix has conducted many scalability tests with Dynamic Memory enabled and found it to be a

very useful and recommended tool. However, there are a few areas to consider during planning:

Dynamic Memory does not oversubscribe the memory pool. When all of the physical

memory has been assigned, used Desktops will not be able to request additional

memory.

The CPU costs to manage Dynamic memory are real, but very reasonable. (<1%

additional CPU overhead).

An increase in IOPS is normal and should be planned for. When a virtual machine

requests additional memory, it will also start to utilize its page file in preparation for its

request being denied by Hyper-V and the operating system running out of physical

memory. Therefore, it should be expected that a virtual machine will increase its I/O

characteristics during this time, which can result in an increase in IOPS of up to 10%.

This is a short term increase, and one noticed more in mass scalability testing than in real

life situations. However, on systems where RAM is more plentiful than IOPS, Dynamic

Memory settings should be conservative. For example if a Windows 7 virtual machine,

with a subset of applications, has been tested and determined to require a minimum of

700MB to run then the Dynamic Memory default setting of 512MB should be increased

to 700MB. By accepting the default setting there will be an increase in IOPS

consumption in relation to paging occurring for the guest.

For more information, including recommended default values for Startup RAM by operating

system, please refer to the Microsoft TechNet Article – Hyper-V Dynamic Memory

Configuration Guide.

Page 17

12. Component Redundancy

The design should ensure that the hardware specifications selected for the infrastructure and

desktops hosts incorporate a sufficient level of redundancy to meet the needs of the business:

Infrastructure Hosts. A component failure on a host supporting the XenDesktop

infrastructure servers could prevent users from connecting to their desktop. Even if

Hyper-V High Availability is enabled, there could be a brief outage while the servers are

restarted on another host.

Desktop Hosts. Component redundancy adds cost to the solution. Therefore, it may

be necessary to tailor the level of redundancy provided according to the requirements of

the user groups. For example, the Executive User group is likely to require a higher level

of redundancy than the HR user group.

For scenarios where a high level of redundancy is required, use redundant storage and network

connections as well as power supplies, fans and disks (typically RAID-1 or RAID-10).

Networking

If unfamiliar with the concepts in this section, please refer to the Microsoft Whitepaper -

Understanding Networking with Hyper-V.

When integrating Hyper-V with XenDesktop it is important to consider the following key

networking topics:

1. Physical Network Connections

If sufficient infrastructure exists, performance may be improved by separating different types of

network traffic across multiple physical Network Interface Cards (NICs), for example

management, cluster, virtual machine, storage, provisioning and backup traffic can all be isolated

from each other. A Hyper-V host used to support a XenDesktop environment typically utilizes

between four and eight physical network connections:

2 x NICs (teamed): Live Migration Network, Cluster Shared Volume Network, Virtual

Machine Network

1 x NIC: Management Network

1 x NIC: Cluster Private Network

If standard network adapters, rather than HBAs, are used to access shared storage, separate out

the storage traffic onto a dedicated network:

2 x NICs (teamed): Storage Network

Page 18

In addition, if the servers are network-constrained and Provisioning Services is utilized it may be beneficial to separate out the provisioning traffic:

2 x NICs (teamed): Provisioning Network

For more information, please refer to CTX120955 - How to Setup a Multi-Homed VM

Separating Large Scale Traffic from ICA Traffic for XenDesktop.

Note: All hosts within a cluster must have the same number of network interfaces and the interfaces must be connected to the same networks.

2. NIC Teaming

The implementation of NIC teaming provides increased resiliency by allowing up to two

network cards to function as a single entity. In the event of one NIC failing, traffic is

automatically routed to the second NIC. The Hyper-V design should determine which network

connections should be teamed for resiliency. Ideally, all network connections should be teamed

apart from the Management and Cluster Networks because they can be failed over to alternative

networks in the event of a failure.

Many servers today are supplied with NICs offering two or more ports. As such, it is important

that any teams created consist of connections from two separate physical NICs so that a single

card failure does not bring down the team.

Redundancy should also encompass the external network. Teamed NICs should be diversely

connected to external switches to help reduce the risk from a single switch failure.

Note: Microsoft does not support NIC teaming with Hyper-V; however, third party solutions are available. All support queries regarding teaming with Hyper-V should be directed to the NIC OEM. In addition, some OEMs may not support TCP Chimney Offload and VMQ functionality with NIC teaming enabled. Please refer to your OEM’s teaming documentation for more information. For more information, please refer to Microsoft Knowledgebase Article KB968703 - Microsoft Support Policy for NIC Teaming with Hyper-V.

3. Virtual Adapter

Hyper-V supports two types of network adapters: Legacy and Synthetic. The synthetic adapter is

better performing, requires reduced host CPU, and is installed by default when a VM is created.

In XenDesktop environments using Provisioning Services the legacy network adapter must be

used in order to PXE boot the desktops. As such, two options must be considered during the

Hyper-V design:

Only use the legacy network adapter, which does not perform as well as the synthetic

adapter.

Configure the Provisioning Services targets with dual network adapters: one synthetic

and one legacy adapter. Once the virtual desktop PXE boots, the synthetic adapter has

Page 19

precedence over the legacy adapter, but PVS traffic is bound to the legacy adapter and is

not able to take advantage of the better performing synthetic adapter.

Note: If the dual NIC configuration is used, each desktop will require two IP addresses from the DHCP server.

The legacy network connection will be suitable for all but the most network intensive

environments. Therefore, for simplicity, utilize the legacy network adapter unless an associated

network bottleneck is identified.

4. Provisioning Services

Provisioning Services is a bandwidth intensive application that can occasionally impact network

performance, particularly if 100Mbps/1Gbps connections are used. Depending on the impact

Provisioning Services will have on the underlying network, it may be necessary to implement a

high-capacity network (10Gbps) or create a separate physical network dedicated to Provisioning

Services traffic. For more information, please refer to the Citrix Blog post – PVS Stream Traffic

Isolation.

For details on how to create a separate Provisioning Services network with XenDesktop, please

refer to CTX120955. For details on the average data transfers during boot-up by operating

system, please refer to CTX125744.

5. IP Addressing

IP addresses need to be assigned to the Hyper-V network interfaces and individual virtual

machines. As such, the design must consider the IP addressing requirements for these

components. If DHCP is used to provide the IP configuration for the Hyper-V hosts, ensure

that reservations are created for the appropriate MAC addresses to prevent configuration issues

when IP addresses change.

The Hyper-V network design should ensure that the Hyper-V traffic is routed via the

appropriate virtual and physical networks. For example, shared storage traffic is routed via the

parent partition and not directly from the virtual machine. Depending on the networking

architecture used, static routes may need to be added to the parent partition’s routing table to

ensure that the correct interface is used.

6. VLANs

Many network environments utilize VLAN technologies to reduce broadcast traffic, enhance

security and to enable complex virtual network configurations which would otherwise not be

possible. The Hyper-V network design should determine which VLANs will be required.

Typical VLANs used in a XenDesktop environment include:

Desktop VLAN

Page 20

Server VLAN

Provisioning VLAN

Separate physical networks are typically created for the Hyper-V Live Migration, Cluster and

Management networks.

Note: Hyper-V supports the configuration and use of 802.1Q tagged VLANs. Each Hyper-V host will need to have its physical NICs connected to specific VLAN trunk ports to allow for the correct routing of VLAN tagged traffic. For more information, please refer to the TechNet Article – Hyper-V: Configure VLANs and VLAN Tagging.

7. Security

For security, firewalls should be used to control traffic flow between the following components:

Virtual desktops and the infrastructure servers

End users and the virtual desktops

End users and the infrastructure servers

The Hyper-V network design should ensure that that port 8100 (WCF) is open between the

XenDesktop Controllers and the VMM server to facilitate machine state queries and power

management operations. For a full list of ports that need to be open for XenDesktop, please

refer to the Citrix Knowledgebase Article CTX101810 – Communication Ports Used By Citrix

Technologies

Note: The Windows Communication Framework (WCF) utilizes Kerberos for authentication and is encrypted

by default. For more information, please refer to the Microsoft TechNet Article – Hardening the VMM server.

Storage

For an introduction to storage, please refer to Citrix Knowledgebase Article CTX118397 –

Introduction to Storage Technologies. For detailed XenDesktop storage recommendations, please

refer to Citrix Knowledgebase Article CTX130632 – Storage Best Practices.

Storage has a major impact on the performance, scalability and availability of the XenDesktop

implementation. As such, the storage design should focus on the following key areas:

1. Direct-attached Storage/Storage-attached network

The Hyper-V storage design must determine whether the virtual machines should be hosted on a

Storage-attached Network (SAN), Direct-attached Storage (DAS) or a combination of the two.

There are a number of factors that need to be considered, including:

Unlike DAS, SAN supports Live Migration, Failover Clustering and Resource

Optimization (PRO). Although these features are less critical when supporting pooled

Page 21

virtual desktops and XenApp servers, they are still very important for dedicated desktops

and infrastructure servers. A SAN must be used if these features are required.

Another disadvantage of local storage is a reduced level of scalability due to the hardware

limit on the number of disks supported, particularly for blade systems. As the number

of virtual desktops per host increases, additional disks are required to accommodate the

number of IOPS generated.

Although using local storage may require additional disks and array controllers to be

purchased per Hyper-V host, the overall cost is likely to be significantly less than that of

an enterprise storage solution.

In many cases, the optimal solution is to use SAN storage for the infrastructure servers and

dedicated desktops so that they can benefit from Failover Clustering and Live Migration while

using less expensive DAS storage for pooled desktops and XenApp servers.

2. Database Storage

If SQL server is to be virtualized, the Hyper-V storage design must allocate DAS or SAN

storage for each of the following databases.

XenDesktop farm XenApp farm EdgeSight for NetScaler

XenApp Power and Capacity Management

Provisioning Services farm

EdgeSight for XenApp / Endpoints

Smart Auditor and Command Center

XenApp Configuration Logging

Virtual Machine Manager

App-V Data Store

3. Tiered Storage

A one-size-fits-all storage solution is unlikely to meet the requirements of most virtual desktop

implementations. The use of storage tiers provides an effective mechanism for offering a range

of different storage options differentiated by performance, scalability, redundancy and cost. In

this way, different virtual workloads with similar storage requirements can be grouped together

and a similar cost model applied. For example, the storage requirements for the Executive user

group are likely to be very different to those of the HR user group.

4. Performance

The performance requirements for each desktop group must be considered during the design of

the storage solution:

Storage Controllers. High performance solutions include Fiber Channel SAN and

hardware iSCSI, while lower throughput can be achieved using standard network

adapters and configuring software iSCSI. The performance of Fiber Channel and iSCSI

can be improved by implementing multiple storage adapters and configuring them for

multipathing, where supported.

Page 22

RAID. Virtual desktops are typically write-intensive:

PVS - 80% write / 20% read

MCS - 50% write / 50% read for MCS

Therefore, the Provisioning Services write-cache and MCS difference disks should be

hosted on RAID-1 or RAID-10 storage in preference to RAID-5 due to the associated

write penalties:

RAID Level Write Penalty

RAID-0 1

RAID-1 & RAID-10 2

RAID-5 (3 data & 1 parity) 4

RAID-5 (4 data & 1 parity | 3 data & 2 parity) 5

RAID-5 (5 data & 1 parity | 4 data & 2 parity) 6

Table 8: IOPS Write Penalty for RAID Configuration

The disk activity for the Provisioning Services vDisk Store will primarily consist of reads,

provided that it’s not used for private vDisks or server side caching. Therefore, RAID-5

offers an acceptable solution at a reduced cost to RAID-10.

Number of VMs per LUN. Storage LUNs that are connected to Hyper-V using Fiber

Channel or iSCSI may be represented as a single iSCSI queue on the storage side (storage

dependent). When a single LUN is shared among multiple virtual machines disks, all I/O

has to serialize through the iSCSI queue and only one virtual disk’s traffic can traverse

the queue at any point in time. This leaves all other virtual disks’ traffic waiting in line.

Although it is possible to parallelize iSCSI commands by implementing active/active

multipathing or optimizing the command traversal using techniques such as command

batching, LUNs and their respective iSCSI queue may still become congested and the

performance of the virtual machines may decrease.

Therefore, the number of virtual desktops allocated per LUN should typically be

restricted to between 20-30 virtual machines. The actual virtual machine per LUN ratio

is environment and workload specific and needs to be determined by means of

performance monitoring (i.e. VM specific read/write latency) and with reference to the

best practices of the storage vendor.

IOPS. The number of IOPS generated will vary based on application set, user behavior,

time of day and operating system used. Scalability testing should be performed to

determine the IOPS required during boot-up, logon, steady state and log-off phases.

The following table provides an estimate on the number of IOPS that will be generated

based on FlexCast model and user performance category:

FlexCast Model Performance Category Steady State IOPS per User

Hosted VM Light 8

Page 23

Normal 15

Heavy 50

Hosted Shared Light 2

Normal 4

Heavy 8

IOPS performance can be improved by selecting an appropriate RAID type, adding disks to the RAID array and/or using high performance disks (e.g. SAS 15k / SSD). For more information, please refer to the Citrix Knowledgebase Article CTX130632 – Storage Best Practices.

Note: The boot process typically generates the highest level of disk I/O activity. As such, virtual

desktops should be started in batches prior to the beginning of the business day to help reduce the load on

the storage subsystem. In addition, disabling the automatic restart of virtual desktops following logoff

will also help to reduce storage load.

Current and future requirements for multiple staggered shifts must be considered, as

there is likely to be a significant impact on performance due to the increased logon and

logoff activity.

Note: The number of IOPS generated by the virtual desktops can be reduced through operating system

optimizations. For more information, please refer to CTX124239 – Optimizing Windows XP for

XenDesktop and CTX127050 – Windows 7 Optimization Guide.

5. Thin Provisioning

Thin provisioning allows more storage space to be presented to the virtual machines than is

actually available on the storage repository.

NTFS has not been optimized for thin provisioned LUNs, specifically when an “over-commit”

state may occur, where the amount of virtual capacity reported exceeds the actual physical

capacity. Microsoft provides a number of recommendations to improve the reliability of NTFS

with thin provisioning that should be considered during the design of Hyper-V storage:

Do not create multiple volumes on a single thinly provisioned LUN

Do not store a system volume on a thinly provisioned LUN

Do not store paging files on a thinly provisioned LUN

Contact your storage vendor for guidance on whether they support thin provisioning with

Windows 2008 Server R2. For more information, please refer to Microsoft Knowledgebase

Article KB959613 – Guidelines for Thinly Provisioned LUNs in Windows Server 2008 R2.

Warning: If using thin provisioning in production environments, take appropriate measures to ensure that

sufficient storage exists and there is a system in place to alert administrators when storage is running low. If

available disk space is exhausted, virtual machines will fail to write to disk, and in some cases may fail to read

from disk, possibly rendering the virtual machine unusable.

Page 24

6. Data De-Duplication

Storage requirements may be reduced through the use of data de-duplication, whereby duplicate

data is replaced with pointers to a single copy of the original item. There are two

implementations of de-duplication available:

Post-Process De-Duplication. The de-duplication is performed after the data has

been written to disk. Post-process de-duplication should be scheduled outside business

hours to ensure that it does not impact system performance. Post Process de-

duplication offers minimal advantages for pooled desktops as the write cache /

difference disk is typically reset on a daily basis.

In-Line De-Duplication. Examines data before it is written to disk so that duplicate

blocks are not stored. The additional checks performed before the data is written to disk

can sometimes cause slow performance. If enabled, in-line duplication should be

carefully monitored to ensure that it is not affecting the performance of the XenDesktop

environment.

Hyper-V Failover Cluster

A Hyper-V failover cluster is a group of Hyper-V hosts (cluster nodes) that work together to

increase the availability of hosted virtual machines. Virtual machines on a failed host are

automatically restarted on another host within the same cluster, provided that sufficient resources

are available.

The following areas should be considered when designing the Hyper-V Failover Cluster design:

1. Virtual Machines to Cluster

The following virtual machines are critical to Hyper-V and XenDesktop are should be hosted on

a failover cluster:

VMM servers

Web Interface servers, License servers

SQL servers

Desktop Controllers

Provisioning Services

XML Brokers

XenApp Data Collectors

Dedicated Desktops

Page 25

Pooled Desktops (with personal vDisk)

Pooled desktops and XenApp servers do not typically need to be hosted on a cluster because

they shouldn’t have user-specific data or applications.

Note: Cluster Shared Volume services depend on Active Directory for authentication and will not start unless

they can successfully contact a Domain Controller. Therefore, always ensure that there is at least one Domain

Controller that is not hosted on a Cluster Shared Volume.

2. Number of Failover Clusters

The number of Hyper-V clusters required to support an implementation of XenDesktop varies

depending on the following key factors:

Infrastructure Nodes. When designing Hyper-V clusters, consider separating the host

server components such as AD, SQL and XenDesktop Controllers from the dedicated

virtual desktops by placing them on different Hyper-V clusters. This will ensure that

dedicated virtual desktop resources are isolated from infrastructure resources for optimal

performance and availability.

Performance. Many businesses will have dedicated desktops that require guaranteed

levels of performance. As such, it is sometimes necessary to create dedicated clusters to

meet the service level agreements associated with these desktops.

Application Set. It may be beneficial to dedicate clusters for large dedicated desktop

groups as they share a common, predictable resource footprint and application behavior.

Alternatively, grouping dedicated desktop groups together based on differing resource

footprints could help to improve desktop density per host. For example, splitting

processor intensive dedicated desktops across several clusters will help to distribute the

impact from processor saturation. If XenApp is used to provide virtualized application

access, it is advisable to separate the XenApp servers onto a separate Hyper-V cluster to

balance resource utilization.

Physical Network. Some environments may have complex network configurations

which require multiple clusters to be deployed. For example, some dedicated desktop

groups may need to be isolated onto specific subnets for reasons of security, whilst

others may have requirements for network connectivity which can only be satisfied

within specific data centers or network locations.

Virtual Network. Depending on the environment, it may be overly complex to trunk

every VLAN to every Hyper-V cluster. As such, it may be necessary to define clusters

based on the VLANs to which they are connected.

Page 26

Processor Architecture. Hyper-V requires that nodes within the same cluster have

processors from the same family. Therefore, the Hyper-V design will need to consider

the processor architecture available when determining the number of clusters required.

3. Nodes per Failover Cluster

Failover Cluster supports up to 16 nodes, however consider reserving at least one node in the

cluster for failover and maintenance.

A single cluster can support up to a maximum of 1,000 virtual machines. It is unlikely that a

cluster hosting dedicated virtual desktops would consist of 16 nodes as this would limit each

node to 63 dedicated virtual machines (66 if one host is dedicated to HA). It is far more likely

that a Hyper-V cluster supporting XenDesktop dedicated virtual desktops will consist of

between 7 and 13 VMs:

7 nodes per cluster - 167 VMs per node with one dedicated to HA

14 nodes per cluster - 76 VMs per node with one dedicated to HA

For more information, please see the Microsoft TechNet article Requirements and Limits for

Virtual Machines and Hyper-V in Windows Server 2008 R2.

Note: When multiple virtual machines exist for each server role, ensure that they are not all hosted on the same

physical host. This will help to ensure that the failure of a single virtualization host does not result in a service

outage. In addition, the physical hosts supporting the core infrastructure components should ideally be located in

different chassis/racks.

4. Cluster Shared Volumes

A Cluster Shared Volume (CSV) allows virtual machines that are distributed across multiple

cluster nodes to access their Virtual Hard Disk (VHD) files at the same time. The clustered

virtual machines can all fail over independently of one another. CSVs are required for Failover

Clustering and Live Migration functionality. Therefore infrastructure servers and dedicated

desktops are typically hosted on CSVs.

The following recommendations should be considered during the Cluster Shared Volume

design:

Microsoft recommends that the CSV communications take place over a different

network to the virtual machine and management traffic.

The network between cluster nodes needs to be low latency to avoid any lag in disk

operations but doesn’t need to be high bandwidth due to the minimal size of metadata

traffic generated under normal circumstances.

Page 27

Note: Since Clustered Shared Volume communication occurs over the Server Message Block (SMB)

protocol, the Client for Microsoft Networks and File and Printer Sharing for Microsoft Networks

services must be enabled on the network adapters used for the cluster network. Disabling NetBIOS over

TCP/IP is recommended.

For more information, please see the Microsoft TechNet article Requirements for Using Cluster

Shared Volumes in a Failover Cluster in Windows Server 2008 R2.

System Center 2012 Virtual Machine Manager

The use of System Center 2012 Virtual Machine Manager (VMM) provides a number of key benefits

to a Hyper-V implementation supporting XenDesktop, including:

1. Simplified integration with XenDesktop

2. Integrated Performance and Resource Optimization (PRO) of Virtual Machines

3. Intelligent Placement of Virtual Machines

4. Central Library of Infrastructure Components

5. Delegated management and access of Virtual Machines

For more benefits, please refer to the Microsoft Article – Virtual Machine Manager Top Ten

Benefits. For more information on VMM, please refer to the Microsoft Whitepaper – Virtual

Machine Manager Technical Documentation.

1. VMM Architecture

There are four key components which make up the VMM role:

1. Management server

2. Database

3. Administrative console

4. VMM library

The VMM components can be installed on either physical or virtual servers. Within

XenDesktop, a single desktop group can be associated with only one VMM server; however one

VMM server can manage multiple desktop groups.

Note: VMM 2012 does not support SQL Server Express as a database option. Only SQL Server is

supported, either Standard or Enterprise edition.

Note: The VMM Administrator Console must be installed on each XenDesktop Controller within the site. In

addition, when the latest Provisioning Services 6.1 hotfixes have been applied, the VMM Administrator Console

must also be installed on the Provisioning Services server so that the Setup Wizards functions correctly.

Page 28

1. VMM Design and Scalability

Design

When designing a VMM 2012 environment there are two design choices that should be

considered. The two designs below are the tested and verified designs according to the

Microsoft Fast Track Reference Architecture Guide.

Design 1 (Recommended): A single VMM virtual machine exists within a host cluster, or a series

of VMM servers each supporting no more than 2000 Virtual Desktops each. These VMM virtual

machine have no VMM application level HA, however they do have VM based high availability

enabled with the Hyper-V cluster. The VM operating system and the VMM application

components can easily migrate between hosts in the cluster in the event of a hardware failure.

This design offers a simpler setup and is the most often used. SCVMM failure will not cause

user connection errors or user disconnects, however it will prevent XenDesktop from starting or

provisioning new desktops. Therefore, short windows of service disruptions are allowable

within the SCVMM design.

Design 2: Two physical servers are used to support VMM 2012 with the High Availability

feature role enabled. These two servers should not be virtualized. This design offers VMM

application layer HA, however it can also increase complexity during the initial setup for the

environment. This design is also more expensive, as each physical cluster of SCVMM server can

still only support up to 2000 Virtual desktops per cluster instance.

Refer to the Microsoft Fast Track Reference Architecture Guide for more information on VMM

2012 architecture guidance.

Scalability

The following considerations should be taken into consideration when determining the number

of VMM servers required:

Citrix has found that the best performance is achieved when each VMM server is limited

to managing 2000 virtual desktops.

While it is possible to run other applications on the VMM server, it is not recommended,

especially other System Center 2012 applications because they tend to have heavy

resource demands and could significantly impact VMM performance.

The following table provides recommended specifications for the key VMM

infrastructure components:

Component Recommended Recommended (>150 Hosts)

CPU Dual-Processor, Dual-core,

2.8 GHz (x64) or greater

Dual-Processor, Dual-core,

3.6 GHz (x64) or greater

Page 29

Memory 4 GB 8 GB

Disk space

(no local DB)

40 GB 50 GB

Disk space

(local DB)

150 GB Use a dedicated SQL server

CPU Dual-processor, Dual-core,

2 GHz (x64) or greater

Dual-processor, Dual-core,

2.8 GHz (x64) or greater

Memory 4 GB 8 GB

Disk space 150 GB 200 GB

CPU Dual-processor, Dual-core,

3.2 GHz (x64) or greater

Dual-processor, Dual-core,

3.2 GHz (x64) or greater

Memory 2 GB 2 GB

Disk space

(no local DB)

Size will depend on what

will be stored

Size will depend on what will

be stored

Page 30

Operating System Delivery

This section compares and contrasts the three main methods of operating system delivery available

with Hyper-V, including Template Installs, Provisioning Services and Machine Creation Services.

1. Templates

Templates are created using a previously installed and configured virtual machine. Templates are

stored within the VMM Library to simplify deployment across all managed hosts. New virtual

machines created from a template receive a unique identity including MAC address, IP address

and computer name. Templates are typically used for ensuring that infrastructure servers (Web

Interface, XenDesktop Controller, Data Collectors, etc.) are created quickly and consistently.

The following recommendations should be followed when designing the templates required for

Hyper-V.

1. Provision space in VMM for a Windows 2008 R2 template to use for building servers in

the infrastructure. This helps ensure server consistency and speeds up the server build

process. Microsoft recommends a disk size of 40GB or greater for a Windows Server

2008 R2 system. Actual disk space required will vary according to the applications and

features that are installed. If the server will have more than 16 GB of RAM, more disk

space will be needed for paging, and dump files.

2. Templates will need to be created for Provisioning Services. Ensure that you allocate

sufficient space for each desktop type required. Separate templates will be required if

different local write-cache sizes or NICs per desktop are required. The Streamed Virtual

Machine Setup Wizard will allow a different amount of vCPUs and RAM to be specified

based on a single template.

3. Use fixed disks instead of dynamic for optimal performance and to ensure that the

Cluster Shared Volume will not run out of disk space as the VHDs expand.

4. It is easy to expand and very difficult to shrink the size of a virtual disk once it has been

created. Therefore, start with a small disk and manually expand it later on if necessary.

2. Machine Creation Services

Machine Creation Services (MCS) does not require additional servers; it utilizes integrated

functionality built into Microsoft Hyper-V. Each pooled desktop has one difference disk and

one identity disk. The difference disk is used to capture any changes made to the master image.

The identity disk is used to store information such as machine name and password. When a

pooled desktop reboots, the difference disk is deleted and the user starts with a brand new

virtual desktop.

IntelliCache is not supported with Hyper-V. As such, MCS generates approximately 50% more

IOPS than Provisioning Services because there is no read or write caching performed.

Page 31

Therefore, MCS is not typically used for large XenDesktop environments hosted on Hyper-V.

However, MCS is still a good solution for small and medium sized XenDesktop deployments

due to the reduced number of infrastructure servers required and a simplified implementation

process. Unlike Provisioning Services, MCS cannot be used to provision XenApp servers. If

XenApp is to be provisioned in addition to virtual desktops, consider standardizing on

Provisioning Services.

If MCS is selected, the Hyper-V design must ensure that sufficient storage space is available:

Snapshot. Each storage repository used for MCS must include space for a full snapshot

of the master image.

Difference Disk. Each virtual machine has a difference disk which is used to store all

writes made to the virtual machine. The different disk is the same size as the base virtual

machine. Thin-provisioning can be used (if supported by the storage solution) to reduce

the amount of total storage required.

Identity Disk. Each virtual machine has an identity disk (16MB).

3. Provisioning Services

Citrix Provisioning Services allows a single virtual hard disk to be streamed simultaneously to

numerous virtual desktops or XenApp servers helping to reduce storage requirements and

simplify image management.

Provisioning Services is significantly more complicated to implement than MCS, however

Provisioning Services generates less IOPS than MCS when implemented with Hyper-V.

Therefore, Provisioning Services is typically used for large enterprise deployments and MCS for

small to medium sized businesses.

The following recommendations should be followed when designing Hyper-V to support

Provisioning Services:

Sufficient storage must exist for the Provisioning Services store to support the estimated

number of vDisks required, including backups and future updates. The vDisk store

should be configured so that Provisioning Services can leverage the Windows System

Cache. This can either be accomplished by hosting the vDisk store on a block-level

storage device with the default Provisioning Services configurations or on CIFS storage

through modifications to the Provisioning Services registry. For more information,

please refer to the Citrix Blog Post – Provisioning Services and CIFS Stores.

Each target provisioned from a shared vDisk must have sufficient storage available to

host its write-cache, which can either be hosted on the target itself (RAM/local

storage/shared storage) or a Provisioning Server (local storage/shared storage). In most

Page 32

situations, consider using either local or shared storage on the target device due to the

following concerns:

o Device-RAM. This can be an expensive use of memory and targets will fail if

memory is exhausted.

o Provisioning Services-Storage. Adds additional latency as requests to/from the

cache must cross the network.

Note: Live Migration, Failover Clustering and Performance & Resource Optimization (PRO) will

not be available when the write-cache is hosted on local disk.

Use fixed-size virtual hard disks for the Provisioning Services write cache because

dynamic disks suffer from a performance hit associated with disk expansion. The virtual

hard disks stored in the Provisioning Services vDisk Store can be either fixed or dynamic

because they rarely change in size and also leverage the Windows Server Cache. For

more information, please refer to the Citrix Blog Post – Fixed or Dynamic vDisks.

The Windows operating system used to host Provisioning Services is able to partially

cache vDisks in memory (system cache). In order to maximize the effectiveness of this

caching process, a Provisioning Server should have sufficient memory available. The

following formula outlines how the minimum amount of memory for a Provisioning

Server can be determined:

System Cache = 512MB + (# of active vDisks * Avg. data read from vDisk)

Total Server RAM = Committed Bytes under load + System Cache

If the amount of data read from a vDisk is unknown and cannot be determined, it is a

common practice to plan for a minimum of 2GB per active Desktop vDisk and 10GB

per active Server vDisk. For more information, please refer to the Citrix Knowledgebase

Article CTX125126 – Advanced Memory and Storage Considerations for Provisioning

Services.

There are five options for storing the cache file of provisioned virtual machines:

1. Cache on Device HD

2. Cache on Device Hard Drive Persisted

3. Cache in Device RAM

4. Cache on a Server Disk

5. Cache on Server Persisted

Page 33

The use of Cache on Device HD is recommended for the majority of XenDesktop and

XenApp implementations because it offers fast performance without consuming

expensive RAM.

Note: Live migration and Failover Clustering is not supported when using Cache on Device HD with

local storage. As such, consider using shared storage for dedicated virtual desktops, XenApp servers and

pooled desktops with Personal vDisks.

4. Desktop Customization

Personal vDisks are available in all versions of XenDesktop 5.6. Generally, the use of a Personal

vDisks is appropriate when there is a strong desire for personalization of the virtual desktop.

This could include a need to use a variety of departmental applications with a small, distinct user

groups or general personalization that is beyond what is available in the user profile. For more

information, please refer to the Citrix Knowledgebase Article CTX133227 – Citrix Personal

vDisk Technology Planning Guide.

The Hyper-V design should consider the following key Personal vDisk design decisions:

The Personal vDisks must utilize one of the storage solutions supported by Hyper-V.

For more information, please refer to the Storage section of this document.

Note: Microsoft does not support Network Attached Storage (SMB, CIFS or NFS protocols) with Hyper-V. For more information, please refer to Microsoft Knowledgebase Article KB2698995 – Microsoft Support Policy for Hyper-V Environments Utilizing Network Attached Storage.

The Personal vDisk consist of two drives; a persistent user disk (PUD) attached to each

VM which contains user profile information (typically the contents in the C:\Users

folder in Windows), and a hidden vhd drive which contains information about installed

applications. By default, both are sized 50% of the total volume size. So when a default

sized 10 GB PvD is created, 5 GB is dedicated to the PUD, and the remaining 5 GB is

assigned to the VHD. If a profile management solution is in place, consider increasing

the size of the application space to 70% or 80% of the total volume size. This change

has to be set in the master image before deploying desktops.

The application deployment strategy chosen can significantly alter the expected

hypervisor scalability. It has been observed that performance and scalability degrade

when more applications are installed into a Personal vDisk. The larger the application

and the more intensive the registry operations it performs, the greater the impact will be

in terms of hypervisor resource utilization. As such, it is important to incorporate an

application deployment strategy that is cohesive with the personalization needs of end-

users. Consider the use of published applications and application streaming when

possible to minimize the impact on scalability. For more information, please refer to the

Citrix Knowledgebase Article CTX126773 – Application Integration for Virtual

Desktops.

Page 34

Note: Scalability testing should be performed to determine the performance impact of Personal vDisks

with the anticipated application set.

Management

Monitoring

Designing and implementing an appropriate monitoring solution for Microsoft Hyper-V will help to

maintain the consistency, reliability and performance of the XenDesktop infrastructure. As such, the

Hyper-V design should include the following key monitoring decisions:

1. Management Tool

It is imperative that the monitoring tool selected includes support for Hyper-V. The use of

Microsoft System Center 2012 Operations Manager with the Hyper-V Management Pack fully

supports Hyper-V 2008 R2 and includes Performance and Resource Optimization (PRO) which

allows virtual machines to be automatically load balanced between nodes within a cluster. For

more information, please refer to the Microsoft TechNet Article – Operations Manager.

2. Management Tool Access

Selected tools should only be made available to relevant personnel. The ability to download,

launch and connect to servers using unauthorized copies of support tools should be minimized

wherever possible. Ensuring that each administrative tier is only available to authorized

personnel will ensure that unauthorized changes, intentional or otherwise, which could affect the

stability of both physical and virtual hosts and the associated cost of resulting systems outage,

can be avoided.

3. Hyper-V Hardware

Vendor supplied tools should be used where possible to capture and analyze bespoke hardware

events.

The high availability feature of Hyper-V can be used to mitigate the impact from a single host

failure. However, there will still be an element of downtime while the virtual machines are

restarted on alternative hosts. Effective proactive monitoring can help to avoid the unnecessary

invocation of high availability and the restart of virtual workloads.

4. Hyper-V Performance

The overall performance of each Hyper-V host will affect the performance of the virtual

machines that it supports. It is critical that the Hyper-V hosts are monitored for bottlenecks

within the processor, memory, disk and network subsystems:

For more information and detailed recommendations, please refer to the Microsoft Blog Post –

Monitoring Hyper-V Performance.

Page 35

5. XenDesktop Infrastructure

It is imperative that the XenDesktop environment is monitored in addition to the Hyper-V

infrastructure. For more information and detailed recommendations, please refer to the Citrix

Knowledgebase Article CTX133540 – Monitoring Citrix Desktop and Datacenter.

Backup and Recovery

The Hyper-V design should consider the backup requirements of the XenDesktop infrastructure.

For detailed information on the backup capabilities of Hyper-V, please refer to the Microsoft

TechNet article – Planning for Backup.

1. Backup Software

To backup virtual machines from within the parent partition the backup application used must

be compatible with both Hyper-V and the Hyper-V Volume Shadow Copy Service (VSS) writer.

For more information, please refer to the Microsoft Knowledgebase Article KB958662 – How

to Backup Hyper-V Virtual Machines from the Parent Partition.

Note: If using Windows Server Backup to backup virtual machines from the parent partition, it is not possible

to backup individual virtual machines. Also, Windows Server Backup does not support the backup of virtual

machines hosted on a Cluster Shared Volume.

2. Backup Type

The Hyper-V design must determine whether the backups will be performed on the host or

within the VM:

Backup from Hyper-V host. Performing the backup from the Hyper-V host is

preferred because it captures more data than performing the backup from within the

guest operating system including the configuration of virtual machines and snapshots

associated with the virtual machines.

Backup within virtual machine. The backup application runs within the virtual

machine. Use this method when backing up data from storage that is not supported by

the Hyper-V VSS writer.

3. Components to Backup

The following Hyper-V and XenDesktop components should be backed up so that it is possible

to recover from a complete failure:

XenDesktop database

XenApp database

Provisioning Services database

Provisioning Services vDisks (virtual desktops and XenApp servers)

Page 36

Hyper-V VMM database

Hyper-V VMM Library

Hyper-V host system state

Hyper-V host local disks

Hyper-V Cluster quorum

Dedicated virtual desktops

Web Interface configuration

License files

User profiles / home folders

Note: It is assumed that there is a fast automated rebuild process in place for the servers (XenDesktop

Controller, XenApp server, Web Interface server, Provisioning Server, etc.). If this assumption is not true, all

infrastructure servers must also be backed up. To simplify the backup and restore process, user data should not be

stored on the virtual desktops or XenApp servers.

It is imperative that the peripheral infrastructure is also backed up – including the streamed

application profiles, Active Directory, file servers (profile and home folder share, etc.) and

application servers.

Virtual networks are not included in a full server backup. Therefore, the Hyper-V design should

document the virtual network configuration so that it can be manually recreated if necessary.

High Availability

Hyper-V Failover Clustering functionality allows specific virtual machines to be restarted on a new

physical host within the same cluster should the current host system fail. For information on how

to configure Hyper-V Failover Clustering, please refer to the Microsoft TechNet Article – Using

Hyper-V and Failover Clustering.

It is important that all components of the infrastructure are considered when designing a highly

available XenDesktop solution:

1. VMM server

If the VMM server is unavailable, XenDesktop will not be able to manage the power state of the

virtual desktops that it manages or create additional virtual desktops. Therefore, Microsoft

Failover Clustering should be included in the design to ensure that the VMM server is highly

available. For more information, please refer to the Microsoft Blog Post – Creating a Highly

Available VMM server.

2. VMM Library

The VMM Library is typically required for the creation of new desktops, XenApp servers and

infrastructure servers. Therefore, the loss of the VMM Library is unlikely to impact the day-to-

Page 37

day activities of a standard XenDesktop deployment. However, if required VMM supports the

use of highly available library shares on a Failover Cluster.

3. SQL Server

The VMM database contains all VMM configuration information. If the VMM database is

unavailable, virtual machines will still be available but the VMM Console will be unavailable.

Therefore, Hyper-V Failover Clustering should be used to ensure that the SQL Server is highly

available.

4. XenDesktop Infrastructure VMs

If sufficient physical capacity exists within the Hyper-V Cluster, key XenDesktop Infrastructure

virtual machines should be configured for Hyper-V failover clustering to ensure that a single

host failure does not affect the availability or performance of the XenDesktop environment,

including:

Web Interface

XenDesktop Controllers

XenApp Controllers

License Server

Provisioning Services

5. Virtual Desktops

The decision as to whether the host servers for the virtual desktops should be made highly

available will depend on the type of desktops that are deployed. It is not typically necessary to

implement Microsoft Failover Clustering for XenApp servers or pooled desktops without

personal vDisks as they do not contain any user specific data or applications.

Hyper-V Failover Clustering is recommended for important dedicated desktops.

Appendix

Appendix A: Optimizations

Performance tuning of Hyper-V should occur at both the hypervisor host level and at the virtual

machine level. Optimizations that can be applied to improve performance include:

Hypervisor Host Optimizations

By default, Windows Server 2008 R2 uses a balanced power plan in Power Options, which

enables energy conservation by scaling the processor performance based on current CPU

Page 38

utilization. On some systems this can degrade server performance and cause issues with

CPU-intensive applications. To avoid this issue, Hyper-V hosts should be set to use the

High Performance power plan.

Power saving settings in the system BIOS should be set to maximum performance, or they

should be disabled to prevent the server from going into a low power state.

Windows Server 2008 R2 and Hyper-V include support for CPU Core Parking, which

consolidates processing onto the fewest number of processor cores and suspends inactive

processor cores. Hyper-V supports core parking by allowing VM threads to be moved

between cores, and putting cores into a low power state when not in use. The only hardware

requirement to take advantage of CPU Core Parking is a system with more than one core. Be

aware that CPU Core Parking can have a negative effect on Hyper-V performance and

should be used with caution in environments where high performance is important. For

more information and details on how to disable CPU Core Parking, please refer to the

Microsoft Blog Post – Tweaking CPU Core Parking.

The Windows XP setup and diskpart utilities create misaligned boot partitions resulting in

additional disk I/O activity. The diskpart utility included with Windows Server 2003 SP1

and Windows 7 addresses this issue. For more information on calculating the correct offset,

please refer to Microsoft Knowledgebase Article KB929491 – Disk performance may be

slower than expected.

Close any virtual machine connection windows that aren’t being used. Leaving connection

windows open when not in use consumes resources that could be utilized elsewhere.

Pooled virtual desktops should not be configured to automatically restart in VMM as this

will conflict with the power operation commands issued by the XenDesktop Controllers.

Close or minimize Hyper-V Manager when not in use. Hyper-V Manager continuously polls

running virtual machines for CPU utilization and uptime information which uses resources.

Enable offload capabilities for the physical network adapter driver in the root partition to

reduce CPU utilization.

If multiple physical network cards are installed on the Hyper-V host, bind device interrupts

for each network card to a single logical processor to improve Hyper-V performance.

Avoid storing system files on drives used for Hyper-V storage.

Virtual Machine Optimizations

Disable any unnecessary services and remove any unnecessary applications. This will reduce

the VM resource consumption.

Page 39

Do not allocate more than one CPU core to VMs unless they will be running multi-threaded

applications. Single threaded applications will not see a performance increase from multiple

cores unless the application is CPU-intensive.

For best performance configure a 1:1 ratio of virtual processors in the guest VM to available

Limit the number of snapshots taken for VMs. When there is a large hierarchy of snapshots,

Hyper-V has to perform a lot of read operations. Hyper-V must check multiple physical disk

files to find the latest version of data which creates a lot of I/O overhead.

Citrix Consulting published whitepapers specifically dedicated to desktop optimizations when

running Windows 7 and Windows XP. Please refer to both guides for more information.

Appendix B: High-Level Implementation Steps

The following are steps required to perform a XenDesktop 5.5 deployment on a Hyper-V

infrastructure running Windows 2008 Server R2. The steps provided do not go into specific details

and should not be used as a step-by-step handbook. This table is meant to aid when planning an

implementation using either Provisioning Services or Machine Creation Services.

Step Task Reference

1. Verify infrastructure pre-requisites. Verify that the following infrastructure components

are in place before building the environment.

Active Directory domain

DNS services

DHCP services

Physical networks and VLANs

Microsoft and Citrix licenses

Microsoft License server

File servers

2. Assess your hardware requirements. Access the processing, storage, and networking

requirements of all the workloads that will run on one

physical server to determine the hardware

requirements for the physical server. For more

information, please see the Microsoft TechNet article

Hardware Considerations for Hyper-V in Windows

Server 2008.

3. Assess your software requirements. Determine the applications that will be required on

the desktops, and which methods will be used to

deliver them. For more information, please see the

Citrix eDocs article Evaluating Application Delivery

Methods.

4. Determine which edition of Windows

Server to use for building Hyper-V, and the

type of installation to perform.

For more information, please see the Microsoft

TechNet article Virtualization Platform Comparison.

Page 40

5. Disable power conservation settings in the

system BIOS and set the Power State in the

Windows Control Panel to High

Performance.

Power state settings in the system BIOS should be set

to maximum performance or disabled to prevent the

server from going into a low power state which will

affect performance. For more information, please see

the Citrix whitepaper XenDesktop and XenApp Best

Practices, and the Microsoft Support article Degraded

overall performance on Windows Server 2008 R2.

6. Install the Hyper-V role on the physical

servers that will run the role.

If the Hyper-V Server 2008 R2 media is used, the

Hyper-V role will be installed automatically. For more

information, please see the Microsoft TechNet article

Hyper-V Planning and Deployment Guide.

7. Set the Hyper-V and Cluster Shared

Volume anti-virus exclusions in your anti-

virus solution.

For more information, please see the Microsoft

article Windows Anti-Virus Exclusion List.

8. Add firewall exclusions for the services

required by Microsoft and Citrix

technologies that will be used to support the

XenDesktop environment.

For more information, please see Citrix whitepaper

Communication ports used by Citrix Technologies

and Microsoft article Network Ports Used by Key

Microsoft Server Products.

9. Join the Hyper-V host servers to the Active

Directory domain.

All Hyper-V hosts that will be part of a cluster must

be in the same Active Directory domain. For more

information, please see the Microsoft article Hyper-V:

Using Hyper-V and Failover Clustering.

10. Attach and configure SAN unless direct

attached storage will be used.

For best performance, dedicate a private network for

storage traffic. Administrators should check their

routing tables and verify that the storage traffic is not

able to route along networks dedicated to Hyper-V

management or VM traffic. Please see the Storage

section earlier in this document for the advantages

and disadvantages of local and shared storage. Shared

storage must be used for Cluster Shared Volumes.

11. Configure Hyper-V networks. Create separate external networks for VM and Hyper-

V management traffic. For more information, please

see the Networking section earlier in the document

for information on the networks required.

12. Install and configure Failover Clustering

and enable Cluster Shared Volumes for

Hyper-V management servers and dedicated

desktops (if they will be used).

For more information, please see the Microsoft

TechNet article Hyper-V: Using Hyper-V and

Failover Clustering, and see Deploying Cluster Shared

Volumes (CSV) in Windows Server 2008 R2 Failover

Clustering.

13. Verify that the Failover Cluster is working

correctly.

Test a planned and unplanned failover to verify that

the cluster is working properly. For more

information, please refer to the Microsoft TechNet

Article Hyper-V: Using Hyper-V and Failover

Clustering.

14. Build a SQL Server or use an existing SQL

server in the domain for VMM,

For more information, please see the Microsoft

articles How To: Install SQL Server 2008 R2 (Setup),

Page 41

XenDesktop, and Provisioning Server

databases. If this server is built as a VM on

the cluster, reconfigure the automatic start

action and make the VM highly available.

and Hyper-V: Using Hyper-V and Failover

Clustering.

Note: that System Center 2012 Virtual Machine Manager

2012 also supports SQL Server version 2008 with SP2.

XenDesktop 5.6 supports SQL Server versions 2008 and

higher. There is a known issue when using SQL Server 2012

with XenDesktop. See Citrix support article Microsoft SQL

2012 Known Issues. Provisioning Services 6.1 supports SQL

Server versions 2008 and 2008 R2.

15. Build a VMM management server as a VM

on the cluster. Reconfigure the automatic

start action and make the VM highly

available.

For more information, please see the Microsoft

articles Installing a VMM Management Server, and

Hyper-V: Using Hyper-V and Failover Clustering.

16. Build Citrix License Server as a VM on the

cluster. Reconfigure the automatic start

action and make the VM highly available.

For more information, please see the Citrix eDocs on

Citrix Licensing 11.10, and Microsoft article Hyper-V:

Using Hyper-V and Failover Clustering.

17. Create VMs to be used for Provisioning

Services. Install the operating system, latest

updates, patches, and install the VMM

Administrator Console on each

Provisioning server.

When installing the VMM Administrator Console, be

sure to select the option Get the latest updates to

Virtual Machine Manager from Microsoft

Update. For more information, please see Microsoft

TechNet article How to Install the VMM Console.

18. Install Provisioning Services on the VMs

created in the previous step. Configure a

Provisioning Services Store and Device

Collections. Reconfigure the automatic start

action and make the VMs highly available.

For more information, please see the Citrix eDocs on

Provisioning Services 6.1, and Microsoft article

Hyper-V: Using Hyper-V and Failover Clustering.

19. Apply optimizations to the Provisioning

Services server.

For more information, please refer to the

Provisioning Services section of the Citrix

Knowledgebase Article CTX132799 – XenDesktop

and XenApp Best Practices.

20. Create VMs to be used for the XenDesktop

Controllers. Install the operating system,

latest updates, patches, and install the VMM

Administrator Console on each controller.

When installing the VMM Administrator Console, be

sure to select the option Get the latest updates to

Virtual Machine Manager from Microsoft

Update. For more information, please see Microsoft

TechNet article How to Install the VMM Console.

21. Launch the XenDesktop installation on the

VM. Select the server-side components:

Controller

Desktop Studio

Desktop Director

For more information, please see Citrix eDocs article

Installing and Removing XenDesktop Server

Components.

22. Configure Desktop Studio for Hyper-

V/VMM.

Once installation completes, Desktop Studio will

open. Select Desktop Deployment and complete the

configuration specifying:

Name of the SQL Server and XenDesktop

database

Page 42

Microsoft Virtualization as the Host type

FQDN of the VMM management server

Domain\administrator account and

password

Name of the Failover Cluster created in step

12.

Cluster Shared Volume if dedicated desktops

will be deployed

Network that will be used.

For more information, please see Citrix article

CTX127578 - How to Configure XenDesktop 5 with

Microsoft Hyper-V and System Center 2012 Virtual

Machine Manager.

23. Build VMs for the application delivery

solution. Install the operating system, latest

updates, and reconfigure the automatic start

action to make the VMs highly available.

Implement methodology for delivering applications

to the desktop. For more information on application

streaming with XenApp, please see the Citrix eDocs

article Publishing Applications for Streaming. For

more information on publishing hosted applications

in XenApp, please see the Ctirix eDocs article

Publishing Resources using the AppCenter.

If using App-V, please see

24. Build Web Interface VMs. Reconfigure the

automatic start action and make the VMs

highly available.

For more information, please see Citrix article

CTX130917 How to Install and Configure Web

Interface for XenDesktop, and Microsoft article

Hyper-V: Using Hyper-V and Failover Clustering.

25. Build Provisioning Services vDisks or

Machine Creation Services Master Images.

If using Provisioning Services prepare a vDisk to use

for provisioning desktops. If using Machine Creation

Services, build a VM to act as the master image. For

more information, please see Citrix eDocs articles

Managing vDisks, and Preparing a Master VM.

26. Configure anti-virus and firewall exclusions

for the virtual desktops.

For more information, please see Citrix article

CTX124185 Provisioning Services Antivirus Best

Practices, Microsoft support article Virus scanning

recommendations for Enterprise computers that are

running currently supported versions of Windows,

and Citrix whitepaper Communication ports used by

Citrix Technologies.

27. Install the XenDesktop Virtual Desktop

Agent on images.

The Virtual Desktop Agent must be installed for

Provisioning Services and Machine Creation Services

deployments. For more information, please see Citrix

eDocs article Installing and Upgrading to

XenDesktop 5.6.

28. Optimize images for XenDesktop. Perform optimizations on images to ensure optimal

desktop operation. For more information, please see

Citrix whitepapers Windows 7 Optimization Guide,

Page 43

Optimizing Windows XP for XenDesktop, and

Logon Optimization Guide – XenApp/XenDesktop.

29. Create VM templates and store in the VMM

Library.

For more information, please see Microsoft TechNet

article Creating Virtual Machine Templates in VMM,

and Citrix blog Create VMs automatically on Hyper-

V 2008 R2 with the PVS Streamed VM Setup Wizard.

30. If using Provisioning Services, add local

Write Cache disk.

Create a new virtual disk, format, and add it as the

Write Cache to the VM template. Detach the local

hard disk with the operating system from the

template.

Note: In Hyper-V, disks are created in 1GB increments so

select a Write Cache size appropriately.

31. Create machine catalogs in Desktop Studio. For more information, please see Citrix eDocs article

To create a new machine catalog.

32. Use Provisioning Services Streamed VM

Setup Wizard to provision desktops in

Hyper-V, or use Desktop Studio to create

Machine Creation Services desktops in

Hyper-V.

For more information, please see Citrix eDocs articles

Using the Streamed VM Setup Wizard, and Machine

Creation Services Primer – Part I.

33. Add machines to catalogs and create

desktop groups in Desktop Studio.

For more information, please see Citrix eDocs article

To create a desktop group.

34. Install and publish applications for desktop

delivery

For more information, please see Citrix eDocs article

Citrix XenApp 6.5.

35. Deploy Citrix Receiver to target devices. Citrix Receiver is the software client that provides

secure delivery of virtual desktops. User an electronic

software delivery solution to deploy Citrix Receiver to

target devices in the organization. For more

information, please see Citrix eDocs article Citrix

Receiver.

36. Test virtual desktops; ensure everything is

working properly before releasing to end

user community

Product Versions

Product Version

XenDesktop 5.x

Hyper-V 2008 R2

System Center 2012 Virtual Machine Manager

2012

Provisioning Services 6.1

Revision History

Revision Change Description Updated By Date

0.1 Document Created Andy Baker & Ed Duncan 6/11/2012

0.2 Document Modified Tony Sanchez 8/22/2012

1.0 Document Released Andy Baker 9/13/2012

About Citrix

Citrix Systems, Inc. (NASDAQ:CTXS) is a leading provider of virtual computing solutions that help companies deliver IT as an on-demand service. Founded in 1989, Citrix combines virtualization, networking, and cloud computing technologies into a full portfolio of products that enable virtual workstyles for users and virtual datacenters for IT. More than 230,000 organizations worldwide rely on Citrix to help them build simpler and more cost-effective IT environments. Citrix partners with over 10,000 companies in more than 100 countries. Annual revenue in 2011 was $2.20 billion.

©2012 Citrix Systems, Inc. All rights reserved. Citrix®, Access Gateway™, Branch Repeater™,

Citrix Repeater™, HDX™, XenServer™, XenApp™, XenDesktop™ and Citrix Delivery Center™

are trademarks of Citrix Systems, Inc. and/or one or more of its subsidiaries, and may be registered

in the United States Patent and Trademark Office and in other countries. All other trademarks and

registered trademarks are property of their respective owners.