drx - Fotolia
Packing for a trip can be a real challenge. If you overpack, then you're forced to lug around a heavy bag of things you're never going to use. If you underpack, you're stuck wearing the same dirty shirt over and over.
Virtual application and VDI capacity planning at the infrastructure level is the same way. If IT professionals overestimate how much processing, memory, storage and network resources they require, they're going to overspend for resources they didn't need. If they underestimate, however, users are going to run into performance issues, and the entire project is likely to fall apart.
To find a happy medium, IT pros should test, test and test again. They also must understand their users, their existing infrastructure, the applications that are in play and more.
What resources should IT consider for capacity planning?
In terms of physical resources for VDI capacity planning, there are four areas IT pros should keep an eye on.
CPU usage is responsible for the number of desktops they can run on each server. When looking at server CPU, IT should consider factors such as acceptable latency levels, the number of connected devices and any concurrent operations.
Next is memory, which IT calculates based on the apps and OSes users work with on their virtual desktops.
Storage, which is the third hardware consideration, comes down to having enough disk space on the servers to support the unpredictable IOPS that come with VDI because of operations such as antivirus scans.
Finally, IT must ensure that its back-end networking hardware can handle peak use times so users do not run into lag issues.
What goes into VDI capacity planning?
Power users who work with graphics-intensive apps and huge files consume a lot more processing, memory, storage and network resources than others, such as task workers who only work with email and web browsers. As a result, IT must keep track of who uses what apps, how they use them and when. Understanding this information will show usage patterns and help IT allocate the right amount of resources. It will also help IT prevent issues such as boot storms, which occur when a large group of users log in to their desktops at the same time.
IT pros also need to consider the VDI products they will use when planning capacity needs. The servers and hypervisor they choose, as well as software, such as antivirus and management tools, can all affect capacity -- not to mention the desktop delivery method. A nonpersistent approach, for example, requires less storage capacity because users cannot save their settings.
It is also important for IT pros to understand how their current infrastructure works before adding virtual desktops. This can reveal patterns of where issues such as slow app loading times arise and what causes them. Analyzing these patterns helps with VDI capacity planning so IT pros can anticipate what will happen if they add or drop users in the future.
What goes into application virtualization capacity planning?
When it comes to virtual applications, IT must consider where the user accesses the app from, whether it be from the office or from a remote location. The further away the user is from the server housing the app, the more bandwidth it takes to deliver it. IT should also know how the user accesses the apps. If users access virtual apps through a WAN instead of the corporate network, for example, bandwidth can also be an issue. The endpoints users work with, from PCs to tablets to smartphones, also make a difference because different devices support different levels of processing, memory and storage capacity that can take the burden off the servers.
Once they have user needs down, IT should examine the actual applications they are virtualizing. How users interact with their apps affects how the apps consume memory and processing power, as well as how much data is saved. IT should set a minimum acceptable service level for each app to make sure performance stays at a certain standard.
On the back end, IT must determine what features they actually need from their application virtualization product because additional features consume additional resources. Some products, such as Microsoft App-V, include reporting services that create more data that IT must store, for example. It's also important to identify database requirements, as well as how to perform backups and disaster recovery and support failovers with app virtualization.
IT should map out the bandwidth, connection paths and WAN link speeds in its network to identify where it might have some spare capacity. In terms of storage, it is critical to set aside enough resources to host and deliver the apps themselves, the data associated with the apps, the analytics data, the configuration files and more.