Considerations for VDI Assessment and Planning
IDC's Server Virtualization Multiclient Study in 2012 showed that VDI implementers saw server hardware, power/cooling costs, and floor space as the top three areas experiencing the most savings through virtualization. Whether these areas are in alignment with your organization’s expectations on the adoption of VDI, many VDI projects fail as the technical team moves from PoC and Pilot stages to Production. With the understanding that a vague test cases and weak TCO assessments can halt the VDI project, it’s important to avoid these pitfalls by establishing a precise planning model first.
Virtual Desktop infrastructure is the practice of hosting a desktop operating system within a virtual machine (VM) running on a centralized server and distributing the resources through use of a hypervisor. Some overarching benefits of VDI include efficiency, reduction in downtime, increased availability, and maintenance performed at the data center.
Most notable VDI solutions on the market are VMWare vSphere, KVM (free), Microsoft Hyper-V and Citrix XenServer. Most of the leading solutions run on whitebox, bare-metal servers. Therefore as long as it’s x86-based (Intel usually), it generally doesn’t matter what brand of server you’re using. This freedom allows IT to add resources without consideration for contract or dependency on vendor roll-outs or pricing structure changes.
Specifics on server hardware will depend on total number of application users, average load time across business units and/or divisions, but designers generally standardize on Dual Xeon™ V3 E5 systems for the most in scalability for VDI. Even VMware ESX or ESXi server, as an example, has a broad compatibility list of such systems. Among them, systems using the Advantech ASMB-923I server board with Dual Xeon E5-2600 v3 processors come with long life component support for 3 to 5 years.
For VDI, there are a lot of moving parts. Let’s take a look at other considerations to assess VDI capacity and performance requirements.
Defined metrics should be used to capture information on utilization of CPU, memory, network and storage. These metrics are important for proper design and sizing of virtual desktop images required for each group of users, and the infrastructure resources to enable them.
Your planning should include the creation of design documents clarifying specifications of VDI components such as hardware, hypervisor, connection broker, gateway and others, so that you can build use cases for accommodating the number of instances and variety of core applications for the functional requirement.
Once you are ready to create a case example for your pilot phase, you will have to ensure that the Virtual Machine (VM) is optimized for VDI so that it makes the best case for showing the performance and savings gains. This step therefore sets a clear test to either prove or disprove the VDI benefit for your enterprise.
Storage
A proper assessment of storage scenario needs using predefined metrics is a must to capture information on utilization. While assessing the amount of storage, a key consideration is analysis of the expected use of stateless and stateful virtual machines (VMs). The number of planned occurrence of stateless, also known as full VMs, as opposed to stateful, also known as assigned VMs, has a big impact.
This will all relate back to the read/write ratio of the applications, so you will want to use storage hardware with Low latency and High IOPS that Intel® Xeon™ High Performance Storage Servers can deliver.
After a VDI pilot project is up-and-running, capacity problems are easier to solve since it is often a matter of adding disks or spindles. However, these changes have their price, so you will want to match any increase to your TCO targets with projected plans to scale.
Conclusions
Among a list of critical aspects involved in a VDI migration, staying with white box Intel Xeon solutions as your default server CPU will ensure that scale-out VDI is just a matter of rack space. This strategy will ensure that TCO won’t be a shifting formula of vendor product and service models that may have future impact on your VDI plan’s TCO.
Yes, assessment and planning take a large portion of work in a DVI project, but in all, identifying performance lag in early phases of the pilot can prevent exceeding budget. Therefore, it is through a model of setting defined metrics and then routinely monitoring them that an IT team can prove a VDI test case for management.
