If IT departments had a crystal ball that could read the future, VDI capacity planning would be simple, but it remains a complex process.
To create an effective plan, it is vital to understand usage and workflow patterns, the existing VDI deployment and how it has evolved and projected workload requirements. Only then can IT predict the physical resources it needs to support its VDI deployment in the short term and the foreseeable future.
Get to know the users
Assess users and how they work. Some users, such as task workers, are relatively lightweight. They use their desktops only for simple operations, such as sending emails or browsing the web. Other users are power users who administer systems, work with massive graphic files or run resource-intensive applications.
IT should note when and where users work, the peripheral devices they use and fluctuations in patterns. To account for events such as boot storms, notice the level of concurrent activity and at what times they occur.
Application usage has a significant effect on resources, such as processing, memory, storage and network infrastructure. Users who work primarily with Microsoft Word require far fewer resources than software engineers running integrated development environments and other services. If IT does not allocate resources properly during VDI capacity planning, productivity suffers.
Understand the VDI deployment
IT must know exactly what hardware and software it has on hand and how users are working with it. If IT already supports a VDI deployment, it should monitor those systems to see where problem areas exist and streamline management practices to immediately improve the deployment.
IT needs reliable projections of where the organization is heading. Archived data can help identify usage patterns. Analyzing this data produces a baseline that can help IT pros better understand future infrastructure requirements, so they can anticipate adding users, implementing new applications or changing work patterns. IT must also determine what levels of service are acceptable for its systems and applications. For example, IT should determine how long users can tolerate a service disruption.
IT should not confuse the organization's projected growth with anticipated infrastructure needs. Organizations can grow but require fewer employees, or increase productivity but decrease resource demands. Tomorrow's workforce could look nothing like what's in place today.
Value the VDI design
Once IT determines what the current and projected needs are, it can assess what technology it actually needs to achieve that level of support. For example, IT decides whether to implement persistent or nonpersistent desktops or both, then it determines the amount of resources each desktop requires -- taking into account CPU, memory and network requirements.
When estimating resources, VDI shops must also consider the servers the desktops run on, the VDI software and hypervisor they use, any antivirus and management software they implement, how users will work with the resources and more. IT should take into account the resource requirements of the OSes running on the desktops and applications. If IT gears VDI capacity planning toward virtualizing any applications, consider those processes as well -- not only in terms of how employees use the applications but also how IT implements the apps. Don't forget drivers, service packs, patches and anything else that might be required to render and support a virtual desktop.
IT must assess the components it needs to balance virtual machine allocation, stand up clusters and failover systems, implement disaster recovery, interface with external systems and services and support any other technologies integrated in the VDI deployment. VDI shops must also make sure to account for the user data that they must maintain and make accessible to employees.
Test your knowledge of VDI and the user management experience
Take this quiz, and see if you're a pro when it comes to Citrix User Profile Management, VMware View Persona Management and some of the other top tools to manage your profile.
Determine the physical resources
The main considerations for VDI capacity planning are CPU, memory, storage and network utilization.
CPU: CPU capabilities determine how many virtual desktops IT can run on each server, based on the expected utilization of each desktop. If half the desktops require 400 MHz of CPU and the other half require 700 MHz, set the CPU usage accordingly for both scenarios, accounting for such issues as the number of concurrent operations, acceptable levels of latency, connected devices, graphics virtualization and any operations that add overhead. Also keep in mind the server itself, including the host OS, hypervisor and VDI software.
Memory: Memory usage calculations depend on the applications and OSes running on the virtual desktops. The process can be particularly tricky when determining how much memory to allocate to each virtual desktop. If the sum total of the virtual memory exceeds the physical memory and the desktops run concurrently, the host OS has to store and retrieve data from a secondary disk, resulting in significant performance degradation.
Storage: VDI shops must ensure that there is enough disk space to handle the deployment and support the necessary IOPS. In a VDI deployment, IOPS patterns are often write-intensive and can be highly unpredictable. Events such as logon storms, boot storms, patching and antivirus scans add to the challenges.
Networking: The unpredictable nature of VDI can also affect the network, so IT must be sure it can handle the projected traffic at peak times. Individuals connecting across WANs are particularly susceptible to bandwidth and latency issues.
The goal is to come up with comprehensive VDI capacity planning that results in the efficient use of components over time, without overallocating or underallocating resources.
How VDI monitoring tools help with capacity planning
How to deal with data capacity issues
Flash-based storage could be the right move for VDI