At Briforum Boston 2016 Jarian Gibson and myself presented about the Secret Sauce about Outlook Performance in a VDI Deployment”. This topic is multifaceted spanning everything from your Exchange and Outlook configuration to your VDI deployment configuration. So in the spirit of VDI this post will cover some of the vital aspects of your VDI design that will ensure a successful VDI deployment.
So I think back several years when VDI was new to organizations, but the backend infrastructure options required to successfully deploy VDI were very expensive or were so new and expensive that organizations couldn’t consider deploying them. The technologies I am referring to are disk options such as SSD’s and flash, hyperconverged infrastructure and GPUs. I think back to a design that I implemented during this time frame that was well architected, but without the appropriate infrastructure IOPS it took some extreme creativeness to ensure this environment performed.
Here is the situation:
We were deploying a virtualized environment needed to support 3200 non-persistent desktops for nurses and physicians that move around the hospital routinely.
Staff also requires:
- Both Internal and External access to desktops
- Any mobile device supported
The Infrastructure included the following all deployed to traditional SAN disk:
- 4 hosts for the underlying server hypervisor
- 45 + virtual servers for the overall design
- Virtualized applications and desktops
- 15 additional virtual servers for supportive VDI infrastructure
- Deployed on Traditional Disk without GPUs due to cost disadvantages to deploy correctly
- Slow application launching
- Users average login time was about 2 min
- Video performance inconsistency
- Video pauses and pixeling
The resolution came from implementing 3rd party products to better support profile management, and by leveraging additional software to identify other areas of improvement such as network latency. However, overtime the user expectations outgrew the environment once again.
Jumping ahead to Today:
The best way to deploy your VDI environment is to research your GPU options and to evaluate Hyperconverged infrastructure and/or flash options for the SAN.
It is important to consider GPUs for all graphical workloads. This is important because the more the GPU can handle within these requests the more likely your environmental users will have work experience within their virtual workstation that they need to be successful within their workday. GPU’s will ensure that the graphics processing is offloaded which will improve overall application performance. Both your frames per second and pixels per second will perform appropriately, and lastly your design can benefit from GPU Floating Point Operations per second (GFLOPS). GFLOPS will improve the number crunching ability of a graphics processor and can be compared to the million instructions per second (MIPS) that a CPU can deliver.
The choices made when architecting your backend storage can make or break your VDI deployment. At the end of the day it comes down to adequate IOPS to improve performance. Traditional disk may be able to keep up in smaller deployments, but as your deployment grows your disk IOPS will degrade without the appropriate backend infrastructure. Here a couple of options to strongly consider.
Hyperconverged Infrastructure Benefits
- Improved performance through IOPS reduction
- Simplified implementation and management (Turnkey)
- Increased automation of management of resources across nodes
Flash Storage Benefits
- Cost point is flash storage is dropping making this a more affordable option to consider
- Gain extremely fast performance that is scalable
- Obtain more than adequate performance through IOPS
Deploying VDI on the correct backend infrastructure will allow your deployment to more than adequately meet the needs of your organization for the long run. Stay tuned for more information on the aspects of your Exchange design that will ensure that Outlook will perform correctly within your VDI deployment.