I addressed the answers to these (and other) questions in a recent ESG Brief, Server Virtualization: Beyond IT-owned Workloads. There has also been plenty of industryspeak on reference architectures, converged infrastructures, and coalitions, but what is the real value?

Let’s use the recent Cisco & Citrix solution as an example. I recently spoke with a VP of Emerging Solutions and her excitement for such a solution centered on:

1. A turn-key solution that helps avoid risk and complexity and streamlines the procurement process.
2. A single support model that expedites the troubleshooting process and takes the burden off her team.

She also fully understood the value of desktop virtualization and recognized the urgency to extend its value to a much larger user population.

Cisco has been challenged with positioning the UCS inside the enterprise that has a well entrenched server footprint from IBM, HP, or Dell. The Cisco & Citrix solution is exactly what’s needed to demonstrate an architecture and management role in a greenfield opportunity. Desktop virtualization is an ideal use case: it requires new IT infrastructure, it is typically run on a completely different set of gear inside the data center, and ESG research clearly shows that desktop virtualization is a top IT priority.

The challenge ahead lies with creating a message that resonates with senior IT and business leaders; they are the top advocates of desktop virtualization.

Additionally, it looks like Cisco and Citrix have built the CVD (Cisco Validated Design) with NetApp. Now each of the industry leaders has to educate the channel, its partners, and customers on the value of the combination of all three solutions; how it helps drive costs down; how it improves operational efficiency; how it enhances security; and how it clearly differentiates itself from the competition.

This is the first example I have seen of an integrated infrastructure that has the potential to resonate with IT professionals, demonstrate the value of Cisco UCS where it is not necessarily going head to head with the incumbent, and create a great opportunity for Citrix to extend its thought leadership position in a market that is about to be set ablaze.

Read more of Mark’s blog entries at Liquefying IT.

via VMware Upgrades Its Virtualized Desktop Lineup – Virtualization from eWeek.

ESG sees a giant opportunity for companies to help match users with technology, choose the ideal desktop virtualization delivery model (BTW it won’t always be VDI), choose the technology vendor(s), and then monitor the results (which include the end-user experience). Companies such as Liquidware Labs have already established a relationship with Dell Services to help address this exact scenario. These companies also see the value of being engaged early in the game and continuing to provide value as users transition and desktop virtualization becomes embraced by end-users and IT operations.

Tactical VDI deployments should not be considered successful desktop virtualization implementations. It takes a clear understanding of the current end-user community, their jobs and responsibilities, application usage, endpoint choices, and the right technology to successfully build desktop virtualization into your company’s desktop strategy.

Read more of Mark’s blog entries at Liquefying IT.

Server Virtualization

In the future, business decisions will drive application choices and policies. The application will provide its requirements to a highly virtualized platform and the platform will take over and perform many of the manual tasks handled by IT administrators today. Compute capacity will be distributed across large hardware resource pools connected by network connections of different speeds and protocols, all tied to storage and data that spans different storage systems on different continents. Protection, security, and performance a will be policy-driven based on the data. In order to optimize cost and utilization, data is moved from place to place based on lifecycle, requirements, and usefulness of data itself—enabling completely liquid IT.

This all may seem too good to be true, but look at what server virtualization has already done in a relatively short time. It has introduced efficiencies and optimization that had been cumbersome and costly in the past. For example, ESG consistently sees business continuity and disaster recovery as a primary driver to server virtualization implementations. IT has already begun to recognize that sever virtualization extends well beyond server consolidation and improved resource utilization.

With the great success server virtualization has experienced, a massive opportunity still lies ahead:

1. Tier-1 applications. It has been relatively simple for IT for virtualize the first 20% to 30% of applications, but the next 5% remains a challenge. Now is the time for IT operations to work with the “server huggers” to build confidence in the virtualized platform. Technology vendors will have to arm application and line of business owners with best practices, blueprints, and frameworks that are application-specific and demonstrate the value of a virtualized platform.  Expect to see solution purchasing trends that are also application-specific, are massively scalable, and are designed and delivered on a virtualized platform.
2. Management and automation. Data center orchestration and management will play a monumental role in virtualization investments as companies drive toward 100% virtualized environments and begin to build applications specifically targeted for the platform. Microsoft, for example, is making a giant push with System Center messaging and aligning with key technology partners, indicating a clear focus on System Center as the management solution for physical and virtual environments.  Now that virtualized environments have successfully matured, IBM, CA, HP, and BMC are all upping their game as well.
3. Multi-hypervisor deployments. Just as IT has managed and maintained a heterogeneous operating system environment, evidence also suggests that multiple hypervisors will be deployed. IT is heterogeneous by nature, so this makes complete sense. All hypervisors essentially do the same job, but licensing fees for proprietary products are driving enterprises to adopt a second free product. The increasing maturity of Citrix XenServer and Microsoft Hyper-V are grabbing enterprise attention, but VMware is not necessarily being displaced where it exists.
4. Platform affinity. For the last few years, the virtualization market has been ripe with a steady stream of features that complement the core virtualization platform. At this point, the technology has actually surpassed deployments in many cases. Expect to see virtualization vendors continue to share their visionary strategic roadmaps with the end-user community in an effort to shift more applications over to the virtualized platform and establish a strong hold inside the data center. Keep a keen eye on how each vendor will approach the situation: VMware with its core vSphere platform, Microsoft with System Center, Oracle from an application perspective, Citrix with a data center to desktop strategy, and Cisco in the areas of unified platforms and data center networking.

Higher level IT priorities are also helping accelerate the migration to a virtualized platform via:

• Centralization of core IT assets, IT personnel, and applications.
• Consolidation of services, operations, resources, and infrastructure.
• Shared services of resources, professional services, and infrastructure.
• Budget and cost control driven by cost reduction, avoidance, and activity-based pricing.
• Security highlighted with improved safeguards, enterprise policies, data protection, and mitigation of insider threats.

The Bigger Truth

The success of virtualization can be measured by the fact that business decision-makers have worked it into their conversations and become proponents of the technology. In many cases, it is a top IT priority due to its ability to save time, reduce capital expenditures, and greatly improve overall data center efficiency. However, challenges lie ahead as IT works to take the initial success of virtualization and transition it to more business production workloads. The interesting fact is that the transition will have little to do with technology and will instead be heavily weighted toward IT and business process.

Application owners and ISVs largely hold the keys to the next wave of server virtualization success. Application choices, migrations, upgrades, and hardware refresh cycles will act as catalysts for change and potential opportunities to move workloads to the virtual platform. New application deployments will also drive deployment choice. ISVs that see the value in virtualization will transparently build in server virtualization for improved availability, simplified maintenance, and additional value to their customers.

Investing in server virtualization should be a no-brainer for a certain class of predictable workloads—IT often has in-depth knowledge regarding these applications and their performance characteristics. Constant analysis is essential to maintaining momentum as organizations target the next wave of applications suitable for virtualizing. Additionally, management tools that provide visibility, performance metrics, and modeling will help tune investments in virtualization and drive increased confidence for future applications.

The Cloud

Cloud computing—everyone wants it, but no one knows what it is. Primarily driven by its ability to provide improved operational efficiencies, IT agility, and risk mitigation, cloud computing as a platform choice that can be applied across multiple IT initiatives will be top of mind for IT executives. The interesting dynamic that will evolve over time is the perceived risk it poses to current IT operations. ESG speaks with IT operations professionals who clearly see the benefits of cloud computing, but are fearful of the potential security risk it poses and even its impact on headcount reduction, shifting application responsibility, and automating current IT responsibilities. Expect to see IT “cloud teams” that are tasked to define and identify initial opportunities, and map cloud computing to IT initiatives. Some CIOs are already being compensated for migrations to cloud.

As businesses plan to transition virtualization investments to a cloud computing platform, they will be focused on:

• Consumption and delivery. Cloud computing is a self-service consumption model where workloads are deployed and transparently executed either internally or over the Internet and delivered to businesses that only pay for what they consume. IT must work to develop a deployment strategy or application owners and developers will turn directly to the public cloud and risk security breaches, violation of compliance mandates, and data loss.
• A cloud readiness checklist that includes virtualizing X86 based workloads, automating routine IT tasks, and delivering IT as a service on a highly optimized platform. Companies will also need to include deployment of a consumer interface, a chargeback model, and favorable economic consumption models for application and line of business owners. The virtualization platform will need to be highly secure with guaranteed service levels for all deployed workloads.
• An integrated platform that includes servers, networking, and storage.  The consumption model is changing as intercompany alliances are founded to build shared services layers. Cisco, EMC, and VMware have made significant movement in the market with VBlocks, same for HP with its Converged Infrastructure and IBM Cloudburst, and this trend should continue. Cloud maturity’s will rely on a stable, reliable, and scalable platform that is well integrated and ready to quickly adapt to business requirements. The modularity of the platforms will also be important for independent scaling of compute, network, and storage capacity.
• Third party capacity that can be leveraged on demand for additional compute or storage requirements. By no means is IT about to move entire data centers to Amazon or Google, but they will begin to leverage service providers as an extension of their cloud computing infrastructures. Savvis, Terremark, and Bluelock are all going to market with IaaS (infrastructure-as-a-service) offerings. Expect to see the service provider market heat up with additional IaaS opportunities focused on data protection, test and development, elastic capacity, and hosted desktop virtualization. The biggest barrier to adoption? Security and auditability in multi-tenant environments.
• Applications. The majority of focus to date has been on keeping existing applications up and running on an improved platform. This is where server virtualization has made giant strides. The next area of focus is new application design and development. Application owners and development teams are very interested in designing new applications to take advantage of HTML 5 as well as Java and .NET platforms that deliver design and deployment choices far superior to those that exist today. VMware has jumped in the game with the acquisition of SpringSource, Microsoft continues to evolve AZURE, and Amazon and Google are rounding out their services.

Cloud computing is an excellent way to capture the attention of IT professionals, but technology vendors, resellers, and service providers have to be armed with clear strategic roadmaps outlining where they are today and how to ultimately deliver IT as a service. A new skill set and a new language will need to be established for vendors and service providers to learn, become efficient at communicating, and quickly demonstrate the value cloud computing.

The Bigger Truth

The buildup around cloud today is primarily being driven by large technology vendors sharing their strategic directions regarding the future of IT. As vendors and the media build hype around the cloud, there will certainly be benefits to IT, but we may not even end up calling it “cloud.” It doesn’t matter. What does matter is that IT will begin consuming a more modular infrastructure that is likely to be delivered pre-integrated and the consumption of IT will be greatly simplified and economically favorable to the business. Server virtualization will lead IT consolidation efforts, but it won’t stretch much beyond helping with existing x86 workloads. IT will still have to maintain mainframe, UNIX, and other proprietary platforms.

IT operations and application owners will view cloud differently. IT operations will look at cloud as the potential to provide previously unavailable capacity; application owners and developers will look at the cloud as a development and potential deployment platform. Both paths lead to success if managed much in the same way we have always managed IT: by applying the required security policies, maintaining application availability, and driving application productivity. Cost, risk, and time to delivery will always be factors and IT will need to manage these variables to make informed decisions that drive value to the business. No one is suggesting going in on Monday morning and moving an entire data center to the cloud, but the opportunity to move pieces, certain workloads, and even desktop is very real and available.

Background

Server virtualization is a strategic initiative for IT organizations around the world.  ESG conducted an in-depth survey of senior IT professionals concerning their organizations’ most important IT priorities for the next 12-18 months;[1] as shown in Figure 1, server virtualization is clearly the number one priority.

It’s no surprise that server virtualization is a top priority given the dramatic benefits achieved by early adopters.  As more and more applications are deployed onto a consolidated pool of servers, better utilization is slashing the capital costs of equipment.  Operational costs are dropping as well. A consolidated pool of servers reduces the ongoing cost of power, cooling, and space in the data center.  The manpower costs associated with deploying a new application drops as IT managers take advantage of centralized and automated management tools.

Besides the obvious benefits associated with increasing the efficiency of servers in the data center, server virtualization can also be used to enhance the flexibility and availability of IT applications and services.   As applications are freed from the confines of a physical server infrastructure, server virtualization can be used to move those applications between servers and protect users from outages due to a physical server failure. In fact, ESG research indicates that virtual machine disaster recovery, backup, and mobility follow closely behind consolidating more physical servers in the list of the top ten server virtualization initiatives.[2]

Introducing Hyper-V R2

Microsoft Hyper-V R2 is a bare metal hypervisor that enables the hosting of multiple virtual machines on the same physical server. The supported virtual machines can be a mix of almost all Microsoft (server and desktop) platforms in addition to a couple different Linux platforms. Hyper-V R2 is available in the Windows 2008 Server R2 operating system as well as Microsoft Hyper-V Server 2008 R2. Using familiar interfaces and wizards, Hyper-V R2 lets companies take advantage of existing Microsoft skill sets, training programs, and certifications.

Hyper-V R2, which was released in September 2009, includes a number of valuable enhancements to Microsoft’s server virtualization platform.  Performance enhancements include improved performance for virtualized applications which rely on virtual hard disks.  New capabilities including Live Migration and Clustered Storage Volumes were added as well.   As shown in Figure 2, Live Migration enhances the mobility of virtual machines while clustered shared volumes enhance availability during failover clustering.

Live Migration, which was introduced in Hyper-V R2, differs from Quick Migration, introduced in Hyper-V R1.  A quick migration of a virtual machine from one physical server to another incurs a brief period of server unavailability. This gap, which typically last a few seconds, may be felt by end-users and in some cases may require an application restart. Live Migration eliminates this possibility—running virtual machines are moved between physical servers with no interruption.

A Clustered Storage Volume (CSV) is a highly available, shared storage volume that extends the capabilities of a Failover Cluster.  A Failover Cluster is a collection of computers (nodes) that work together to increase the availability of an application or service.  CSV allows multiple servers in a cluster to read and write at the same time while not having one node responsible for the volume’s management.   When a Live Migration is executed and its storage is provided by a CSV, the migration can occur very quickly since only the memory of the virtual machine needs to be migrated.

ESG Lab Validation

ESG Lab performed a hands-on evaluation of Hyper-V R2 server virtualization technology in a lab located on Microsoft’s Redmond, Washington campus.  Testing was performed with a goal of evaluating the flexibility, availability, and performance enhancements introduced in Hyper-V R2.

Getting Started

The test bed used during the ESG Lab validation was designed to emulate a physical to virtual server migration, as shown in Figure 3. Commonly deployed applications and services running on nine older physical servers were consolidated onto a cluster of new servers attached to a shared pool of SAS drives within a FC attached disk array.[3] The servers were configured to support a mix of services and applications including Active Directory and print services, network shared drives (Corp, Dept, Home), Intranet(IIS),  Red Hat Linux(RHEL), and databases (Microsoft SQL Server 2005 and 2008).  Microsoft Windows 2008 R2 Datacenter Edition, with built-in Hyper-V R2 support, was used to consolidate virtual servers onto a Failover Cluster using SAN-attached storage and Clustered Shared Volumes.  A fourth server was used to manage the physical and virtual environment with Microsoft System Center Virtual Machine Manager (SCVMM). [4]

ESG Lab Testing

ESG Lab testing began with a bare metal installation of Windows 2008 R2 on the first 64 bit server.  Six mouse clicks and 20 minutes later, the familiar Windows installation process was complete.  At this point, the server looked and behaved like any physical server running a Microsoft Windows operating system.

The wizard shown in Figure 4 was used to enable the Hyper-V role from the Server Manager console on the physical server.   The wizard-driven user interface used to enable the Hyper-V role was similar to the one used for configuring common server roles including Active Directory and print services.

Six mouse clicks and a reboot later, the Hyper-V R2 enabled server was ready for the configuration of the first virtual machine.

At a high level, three methods are commonly used to create a virtual machine (VM) in a Hyper-V environment:

1. Create a new VM from scratch
2. Clone an existing VM
3. Import an existing physical or virtual server

A wizard launched from the Hyper-V Manager console on the physical server was used to create the first VM from scratch.  Clone and import methods were used later in the ESG Lab validation to create additional virtual machines. As shown in Figure 5, a Hyper-V Manager wizard accessed from the Server Manager console was used to configure the first virtual machine with 512 MB of virtual memory, a virtual network connection, and a virtual hard disk (VHD).

Ten mouse clicks and 42 seconds after getting started with the wizard, the first virtual machine was running a network-based operating system install.

ESG Lab found that getting started with Hyper-V R2 is a straightforward, intuitive process.   At a high level, familiar wizard-driven graphical user interfaces built into Windows Server 2008 R2 were used to configure Hyper-V in three steps:

1. Install Windows Server 2008 R2
2. Enable Hyper-V role
3. Reboot and configure virtual machines

As shown in Figure 6, the first virtual machine was up and running 25 minutes and 22 mouse clicks after starting a bare metal installation of Windows Server 2008 R2.

Availability

Microsoft has improved the availability of applications running within a virtual server infrastructure with Hyper-V R2 enhancements, including a new feature called Clustered Storage Volume (CSV). A CSV can be created and mounted by servers that are attached to the same storage system through a storage area network. By simply turning on the failover and clustering feature on two or more nodes in a cluster, CSV provides automated and transparent failover of virtual machine from one physical server to another. CSV can also be used to accelerate the migration of virtual machines within a cluster.   While CSV creates a single name space which enhances the availability of a Microsoft-enabled virtual server infrastructure, it is recommended but not required for transparent failover or migration.

As shown in Figure 7, cluster shared volume support in Hyper-V R2 enables transparent failover. In this example, a virtual machine running on the failed server on the left has been configured to use a SAN-attached clustered shared volume, which is visible to both of the servers in the cluster. When the failure occurs, running applications fail over transparently to the healthy server on the right.

ESG Lab Testing

ESG Lab tested a CSV-enabled failover using a Microsoft SQL Server 2008 database running within a virtual machine hosted by Hyper-V R2. The database was installed on a clustered shared volume running within a three-node Hyper-V R2 cluster. A SQL select statement was run against a 952,576 row production database harvested from ESG’s internal IT operation.  The long running SQL query performed a join of three tables. The server running the SQL query was powered down as the SQL query was running.

After a few seconds, a remote desktop session connection with the virtual machine was lost.  The SCVMM console was updated to show that the server was no longer available. Forty seconds later, the remote desktop session reconnected automatically with the virtual machine running on a healthy virtual machine in the cluster.

Figure 8 provides an overview of the failover process. The “before” picture on the left shows the state of the cluster before the server was powered down.  At this point, the virtual machine was running on the third node in the cluster (WS08R2-N3). The “after” picture on the right shows the state of the cluster after the failover. Note that SCVMM has changed the status of the cluster and the third node to show a yellow alert.

The performance graph shown in the middle of Figure 8 was captured on the virtual machine running the SQL query during the failover. Note the spike in CPU activity and drop in network activity during the actual failover. Less than three minutes after injecting the error, the virtual machine had automatically failed over to a healthy node in the cluster.  The SQL query that was started before the server failure completed without error.

Flexibility

Live Migration is one of the highlights of Hyper-V R2.  As shown in Figure 9, Live Migration is used to move a running virtual machine from one physical server to another with zero downtime.

Live Migration differs from Quick Migration, introduced in Hyper-V R1, because Quick Migration causes a brief period of application and network interruption as the virtual machine is paused and restarted on a new server. This pause, which lasts less than a minute and can be tolerated by most applications, has been eliminated with Live Migration.

ESG Lab Testing

Selecting which virtual machine to migrate, and which server to move it to, was intuitively simple using Microsoft System Center Virtual Machine Manager (SCVMM). The SCVMM GUI was used to drag and drop virtual machines from one server to another. A right click on a running virtual machine provided yet another intuitive way to kick off a Live Migration. This method is shown in Figure 11.

ESG Lab noted that that SCVMM is not required to perform a live migration.  While the Failover Cluster Manager snap in that is built into Windows Server 2008 R2 can also be used to perform a Live Migration,  ESG Lab found that performing and monitoring a Live Migration from SCVMM  is slightly more intuitive and user friendly.

As shown in Figure 11, a Live Migration with an active SQL query running took one minute and eighteen seconds to complete. The SQL query took a bit longer to run than before the migration (17%), but continued to run without interruption. An active remote desktop connection to the virtual machine remained connected throughout.

Note that since the SQL database was installed on a clustered volume that was shared by all of the virtual machines in the cluster, the migration did not require the movement of the database which was stored on a virtual hard disk.  In this example, a copy of a 40 GB virtual C: drive was avoided as Hyper-V R2 moved a virtual machine in less than two minutes.

Scaling Out

As shown in Figure 12, ESG Lab continued to populate a failover cluster (R2-Cluster) deployed on three 64 bit servers to run nine virtual machines supporting a mix of 64 and 32 bit operating systems.

Note that the Microsoft SCVMM GUI can not only be used to manage a consolidated cluster of virtual machines hosted by Microsoft Hyper-V hypervisors, it can also be used to manage virtual machines hosted by VMware ESX.

Performance

Hyper-V R2 includes a number of performance enhancements that increase virtual machine performance.  Hyper-V R2 added support for Second Level Address Translation (SLAT), which uses hypervisor offload capabilities in modern CPUs to improve VM performance. The maximum number of logical processors has increased to 64 and supporting up to 384 virtual machines per server. Live Migration across different CPU versions within the same processor family has also been added (e.g., Intel Core 2-to-Intel Pentium 4 or AMD Opteron-to-AMD Athlon).

Hyper-V R2 also improves the performance of virtualized applications that store information on a Virtual Hard Disk (VHD).  A virtual hard disk is a file that encapsulates a hard disk image. VHDs first were created to be the storage media for virtual machines.  Today, VHDs are used to ship trial versions of software, in backup solutions, for bug triage (e.g., customers can convert a physical disk to virtual and share it), and even to store multiple boot environments.  VHD was integrated into Windows Server 2008 R2 which added native support.  Native support means the technology has been integrated into the operating system.

ESG Lab Testing

ESG Lab performed a series of tests designed to confirm Microsoft’s claim that fixed VHD performance is on par with native physical drive performance in Windows Server 2008 R2. The performance of real-world applications before and after virtualization on a modern CPU with SLAT support was tested as well.  Performance testing began with an evaluation of fixed VHD performance on a high performance 64 bit server with 110 GB of RAM and four AMD Opteron 8439 processors with Rapid Virtualization Indexing (RVI is AMD’s version of SLAT), each with six cores (24 cores total).

A fixed VHD is a file that is equal in size to a logical drive when it is created.  Because the physical storage required for a fixed size VHD is allocated when the file is created, there is a better chance at optimal placement and organization on-disk which yields the best performance.  The disadvantage is that space is committed even if it is not used. A dynamically expanding VHD allocates capacity on demand as data is written by applications. While the performance of dynamically expanding VHDs has improved significantly in Windows Server 2008 R2, their read/write performance is generally slower than fixed disks.[6] ESG Lab tested the performance of a fixed VHD.

The industry standard open source Iometer utility was used to generate workloads which mimic the IO profiles of real-world applications typically deployed in a virtual server environment.  Iometer version 2008.06.18 was used. [7] Four workloads were tested:  SQL log, OLTP, file server, and Exchange DB.[8]

Native testing on a server running Windows Server 2008 R2 Datacenter was performed using a 40 GB physical disk (a.k.a., raw disk or bare metal disk). Results were compared with a 40 GB fixed VHD running within a Hyper-V R2 VM configured on the same physical server.  Capacity was provided by a Fibre Channel attached HP MSA storage array with six 15K SAS drives configured in a RAID-5 group (5+1).  A two minute Iometer test run with a five minute ramp period was used. The average of three test runs was compared. The results are shown in Figure 13.

What the Numbers Mean

• ESG Lab testing indicates that native VHD performance is nearly the same as a native physical disk in Windows Server 2008 R2 (95% to 99%).
• Excellent VHD performance is due to a number of performance improvements in Windows Server 2008 R2 including integration of VHD support in the operating system.
• Prior to Hyper-V R2, customers often relied on pass through disks to optimize the performance of applications which store information on a SAN attached disk arrays. Pass through disks can also be used to take advantage of advanced data protection and disaster recovery capabilities built into disk arrays (e.g., snapshot sand remote mirroring).   R2 performance improvements make VHDs a more viable alternative to a pass through disk in a SAN attached environment.

During the next set of tests, ESG Lab evaluated the performance overhead of Hyper-V R2 in Microsoft Windows 2008 R2 Datacenter edition.  Physical and virtual performance was compared for the following application workloads:

1. Application install:  a timed installation of Microsoft SQL Server 2008 using a distribution image stored on a network attached shared drive within a private network.
2. Directory level copy:  a timed copy an 860 MB directory with 2,014 files to a temporary directory.  The c:\windows\win32 directory was copied to a temporary directory on the same C: drive.
3. Subsequent copies:  the directory level copy was repeated with much of the IO activity happening in cache. The average of three cached copy operations was recorded.
4. SQL query:  a long running SQL select statement using a 952,756 row production database from ESG’s internal IT operation was timed.  The SQL query performed a join of three tables. The average duration of three select statements was compared.

Comparing physical and virtual performance on the same server was accomplished after a reboot with Hyper-V role enabled. The physical server was tested with all of the available CPU cores (64) and RAM (110 GB). During virtual server testing, the server was configured with a single virtual machine which used four CPU cores and 64 GB of RAM.   Physical and virtual testing was performed with a 40 GB logical C: drive. The virtual configuration was tested with the C: drive configured as an NFTS volume on a fixed VHD. The results are shown in Figure 14.

What the Numbers Mean

• The difference in performance for real-world application workloads is relatively low (2% to 11%)
• Some of the performance difference may be due to the fact that native Windows Server 2008 R2 tests were performed with all of the available CPU and RAM in the physical server (64 cores, 110 GB of RAM). In contrast, the Hyper-V R2 VM tests were performed with the maximum amount supported CPU and RAM for a single Hyper-V R2 virtual machine (8 cores, 64 GB of RAM).
• In ESG Lab’s opinion, the manageably low performance impact of Hyper-V R2 won’t be detected by the vast majority of end-users and applications.

ESG Lab Validation Highlights

• Using familiar Windows-based tools, ESG Lab installed and configured a Windows Server 2008 R2 Datacenter server with bare metal Hyper-V R2 virtualization support in 25 minutes and 22 mouse clicks.
• ESG Lab has confirmed that Clustered Shared Volume and Failover Clustering support in Microsoft Hyper-V R2 can be used to increase the availability of IT services as it provides quick and transparent recovery of virtual machines after a server failure. A remote desktop connection to a VM running a SQL query was restored forty seconds after powering down a physical server in a cluster.  The SQL query completed without error.
• Live Migration was used to move running applications between physical servers with zero downtime.  A remote desktop connection to a VM running a SQL query remained available throughout the migration. The SQL query completed without error.
• Nine virtual machines running a mix of common business applications and a variety of 32 and 64 bit Microsoft Windows operating systems were consolidated onto a cluster of multi-core blade servers.
• Microsoft management tools with a familiar look and feel (Server Manager with the Hyper-V Manager MMC plug-in and SCVMM) were used to deploy, manage, and optimize the virtualized infrastructure.
• ESG Lab measured manageably low fixed VHD and virtualized application overhead in a Microsoft Windows Server 2008 R2 environment (2% to 11%).

Issues to Consider

• While not tested during this ESG Lab Validation, one of the more compelling capabilities of Microsoft’s integrated suite of virtualization management tools is the ability to manage physical and virtual servers— including virtual servers running over a VMware hypervisor—from a single pane of glass.
• Another core capability not tested by ESG Lab is the ability to Live Migrate VMs between servers in the same processor family (Intel Core 2 to Intel Pentium 4).
• Applications requiring more than the 8 CPU cores and 64 GB of RAM supported by Hyper-V R2 are not a good candidate for server virtualization.
• While a fixed VHD performs better than a dynamically allocated VHD, the disadvantage is that space is committed even if it is not used.

The Bigger Truth

Server virtualization is on a rapid and pervasive adoption path.  The cause—and effect—of this is the clear value proposition provided by server virtualization solutions. In a recent global research survey of 706 end-users, ESG found that the impact and benefits of implementing server virtualization are driving organizations to trust server virtualization with some of their most mission critical applications as they recognize that the benefits extend well beyond physical server consolidation and improved resource utilization.

Clearly, server virtualization provides a wide range of benefits, including better server resource utilization; increased consolidation of physical servers; and reduced consumption of floor space, power, and cooling.  In addition, ESG is finding that server virtualization is rapidly becoming an enabler of server management and data center automation. Virtual machines can be quickly provisioned, optimized, and tracked throughout their lifecycles and management tools are beginning to be able to manage physical servers and virtual machines as well as heterogeneous server virtualization solutions.  Disaster recovery and high availability are important—in some cases, they are the primary drivers for running production workloads on a virtualized infrastructure.

ESG Lab has confirmed that Microsoft’s growing family of server virtualization solutions have been built with each of these benefits in mind. Bare metal Hyper-V R2, which is built into Windows Server 2008 R2, provides the underlying virtualization technology that enables server consolidation. Wizard-driven configuration with a familiar Windows look and feel enables rapid server provisioning (less than 30 minutes during ESG Lab testing). A robust family of virtualization enabled management tools ensures that virtual—and physical—servers can be quickly provisioned, optimized, and tracked throughout their lifecycles.  Powerful flexibility and availability capabilities tested by ESG Lab—including Live Migration, Clustered Shared Volumes, and improved performance—are turning Microsoft server virtualization technology into an enabler of server management and data center automation.

Microsoft’s growing suite of management tools (e.g., SCOM and SCVMM) magnifies the value of Hyper-V R2.  With Microsoft System Center, you can manage at the application level regardless of whether applications are deployed on a physical server or a virtual server. You can manage virtual machines running on a mix of hypervisors from Microsoft, VMware, or Xen.

ESG Lab testing has proven that the performance overhead of Hyper-V R2 is manageably low compared to the outstanding benefits of server virtualization.  Moreover, ESG believes that a misdirected focus on the relative differences in performance between competitive hypervisor technologies is a disservice to IT managers. Given the dramatic benefits of server virtualization and the rapid advances in multi-core server technology, a relative difference between technologies that are 90% or more efficient matters little. The overall capabilities of the solution, including how well it works with existing processes and technologies,  matters much more. As an analogy, imagine you were trading in a car that gets 10 miles per gallon for one of two cars that each gets more than 100 miles to the gallon. Both of the new cars will be more than ten times more efficient that your old car. For most car buyers, an efficiency difference of a mile or two per gallon would be a minor consideration compared to price, performance, service, and options.

Speaking of price, ESG Lab believes that the cost of purchasing and owning a server virtualization solution is a vital consideration. Microsoft Hyper-V Server 2008 R2 is a free download, limited in terms of support for total memory, number of processors, and quick migration. Microsoft Windows Server 2008 R2 Standard, Enterprise, and Datacenter editions each include a Hyper-V R2 license for a single physical server as well as licensing for one, four, and unlimited virtual machines, respectively. Considering the additional savings that can be achieved when you leverage existing investments in Microsoft training, certification, and interoperability, the bottom line is simple: IT managers save time and money with the Hyper-V R2 support built into Microsoft Windows Server 2008 R2.

Appendix

[2] Source: ESG Research Report, 2009 Data Center Spending Intentions Survey, March 2009.

[3] Configuration details are listed in the Appendix.

[4] For simplicity, the management server is not shown in this diagram.

[5] Source: ESG Research Report, 2010 IT Spending Intentions Survey, January 2010.

[6] Microsoft indicates that the performance of a dynamic VHD support is approximately 87% of a native hard disk.  http://blogs.technet.com/b/virtualization/archive/2009/07/22/windows-server-2008-r2-hyper-v-server-2008-r2-rtm.aspx

[7] http://sourceforge.net/projects/iometer/

[8] To learn more about ESG Lab Iometer workloads, see http://www.enterprisestrategygroup.com/using-esg-lab-workloads/.

Here is a snippet from the Lab Validation that walks through the methodology:

A Mixed Real-world Storage Benchmark Methodology

While VMmark is well suited for understanding the performance of a mix of applications running on a single server, it was not designed to assess what happens when a mix of applications is run on multiple servers sharing a single storage system. VMmark tends to stress server internals more than it does the storage system. The methodology presented in the balance of this report was designed to stress the storage system more than the servers. Taking a cue from the VMmark methodology, a tile-based concept was used. As shown in Figure 5, each tile is composed of a mixture of four application workloads. Two physical servers, each configured with eight virtual machines, were used to measure performance as the number of active tiles was increased from one to four.

Figure 5. ESG Lab Tile-Based Storage Benchmarking

The difference between the server-focused VMmark benchmarking and storage-focused ESG Lab benchmarking is shown in Figure 6. Note how VMmark testing is performed with a single server, often attached to multiple storage systems. As a matter of fact, the IBM BladeCenter HS22 VMmark results presented earlier in this report were achieved with a pair of IBM System Storage DS4700 arrays. In other words, when vendors publish VMmark results, they make sure there is plenty of storage available so they can record the highest VMmark server score. This provides IT managers with a fair comparison of the performance capabilities of competitive server technologies.

ESG Lab storage-focused benchmarking uses a different approach. Instead of testing with a single server and more than enough storage, multiple servers are attached to a single storage system. Rather than running application level benchmarks which stress the CPU and memory of the server, lower level industry standard benchmarks are used with a goal of measuring the maximum mixed workload capabilities of a single storage system.

Figure 6. Server-focused VMmark vs. Storage-focused ESG Lab Benchmarking

Mixed Workloads

Industry standard benchmarks were used to emulate the IO activity of four common business application workloads:

• E-Mail: The Microsoft Jetstress utility was used to generate e-mail traffic. Similar to the Microsoft LoadSimm utility used in the VMmark benchmark, Jetstress simulates the activity of typical Microsoft Exchange users as they send and read e-mails, make appointments, and manage to-do lists. The Jetstress utility is however a more light-weight utility than LoadSimm. Using the underlying Jet Engine database, Jetstress was designed to focus on storage performance.
• Database: The Orion utility from Oracle was used to generate database traffic. Much like Jetstress, Orion is a lightweight tool that is ideally suited for measuring storage performance. Orion was designed to help administrators understand the performance capabilities of a storage system, either to uncover performance issues or to size a new database installation without having to create and run an Oracle database. Orion is typically used to measure two types of database activity: response-time sensitive online transaction processing (OLTP) and bandwidth sensitive online analytic processing (OLAP).
• Web Server: The industry standard Iometer utility was used to generate web server traffic. The IO definition was composed of random reads of various block sizes. The web server Iometer profile used for this test was originally distributed by Intel, the author of Iometer. Iometer has since become an open source project. Iometer tests were performed on Windows physical drives running over VMware raw mapped devices.
• Scan/read: The Iometer utility was used to generate a single stream of read traffic. Operations that tend to generate this type of large block sequential traffic include scan and index operations, long running data base queries, backup operations, bulk data uploads, and copies. One 256 KB sequential read workload was included in each tile to add a throughput intensive component to the predominantly random IO profile of interactive e-mail, database, and web server applications. As most experienced database and storage administrators have learned, a throughput intensive burst in IO traffic can drag down the performance for interactive applications, causing performance problems for end-users. Adding a few streams of throughput intensive scan/read traffic was used to determine whether interactive performance would remain predictably responsive as the amount of mixed IO utilization increased.

Each of the four workloads ran in parallel, with the Jetstress e-mail test taking the longest to complete (approximately three hours).

Read more of Mark’s blog entries at Liquefying IT.

VMware in particular has been very successful in helping companies take existing x86 Windows and Linux workloads and consolidate/migrate them to an efficient platform that is simple to manage, improves application availability, and buys back IT’s time. Thus far, VMware’s success has primarily been with IT-owned and -managed applications along with a steady stream of maintenance/support renewals. There is still plenty of opportunity in this market, but VMware is being challenged by the likes of Microsoft with Hyper-V and Citrix with XenServer and by cloud providers like RackSpace and Oracle within their own stacks. The world it once ruled now has some company.

But fear not: VMware is on an acquisition tear to address the future of IT and specifically the future of application design and development. VMware CEO Paul Maritz also jumped up on stage during the day one keynote at Google I/O 2010 to show how VMware, with its acquisition of SpringSource, is laying down the accelerator pedal and looking to help steer application development design decisions for enterprise Java developers.

VMware will continue to help companies migrate existing applications to a virtualized platform and is bullishly positioning itself to be part of the roadmap for future application development. You may even imagine an environment where applications are written specifically to run on the “VMware OS.” But, before that happens, there are some key moves that will have to play out:

• VMware has to aggressively expand its community of users and technology partners beyond its existing base of IT operations-focused personnel. Establishing momentum with application developers, DBAs, and CIOs will be critical to gaining popularity, establishing credibility, and qualifying for pole position. Its announcement with Google is a fantastic way to establish this.
• VMware has not made its business model with developers or the future state of application design 100% clear. On the server virtualization side, it is clear, but SpingSource and the other recent acquisitions muddy the waters. There is certainly an opportunity to arm service providers with out-of-the-box offerings that include messaging and, perhaps even more interesting, databases; how VMware is going to profit has not been laid out.
• The tide is beginning to rise as Microsoft amps up Azure to .NET developers and Oracle gains new enthusiasm with Java. VMware is putting itself right in the mix and has the potential to benefit from any shift in application design.

VMware’s success will also be dependent on its ability to help build a bridge between the IT operations and application teams. The two constituencies operated with very different goals and priorities–and VMware can position itself to be an influencer on both fronts. Essentially, VMware must show value on both sides.

It will be important to watch how VMware continues to make investments in the application space, internally balances two distinct objectives, and then goes to market.

Read more of Mark’s blog entries at Liquefying IT.

via Wyse Technology’s new device isn’t your father’s dumb terminal – Citrix, thin clients, virtualisation, Wyse – Techworld.

via Wyse and Citrix Joint Desktop Virtualization Deployments Highlight Optimized Simplicity, Security, ROI and User Experience – Business Wire – Wire – BradentonHerald.com.