How upgrading to Windows Server 2016 impacts VDI
Windows Server 2016 released several weeks ago, and for RDS and VDI, it’s got a lot of great new features going for it.
A couple of my favorites:
- Direct Device Assignment (DDA) allows you to dedicate GPUs to virtual machines.
- New codecs for clearer text.
- Storage Spaces Direct lets you use local storage to create scalable, software defined solutions.
There’s lots more, of course, like improvements for hosting desktops in the cloud.
One of the interesting things about the last few Microsoft OS releases is that the minimum hardware requirements have not changed much:
In 13 years, the minimum CPU hasn’t really changed, other than support for 32-bit CPUs, which was dropped after Server 2008.
If this was the only data you looked at, you might expect the same performance for end users on any of these OS versions when running on the same underlying hardware. We know that doesn’t make sense though –Microsoft continues to cram more new features into each release, adding things like antivirus, UI that utilizes the GPU, Cortana and more.
So, what’s the performance tax for all those new features? What’s the net impact to users?
If you pay close attention to this blog, my colleague Mark wrote about this last October. We didn’t post all the details, but we gave a hint at our initial findings: User Density with Server 2016 RDS hits max user capacity (what we call ‘VSImax’) 15% sooner than Server 2012 R2 on the same hardware. With image tuning, it’s still about 8% lower capacity. If you’re planning an in-place upgrade, it’s something to be aware of.
Mark’s blog was just a teaser, but we wanted to let you know that we’ve taken all the details from our testing and put them together in a white paper.
It’s got lots of additional information about our testing methodology (if you want to replicate our results in your own lab) and the tuning that went in to improving the results.
You can download the paper here. Use it to start planning your own Server 2016 evaluation. Show it to your boss when you need data to justify any new hardware purchases before upgrading.
And by all means… do your own testing. You wouldn’t buy a car after just reading the window sticker specs – you’d want to test drive it. If you’re like me, you hope the sales guy doesn’t come with you, so you can really see what the car is capable of. Why not do the same with your remote desktops?
Want to know more about our OS version results? Check out our webinar on the topic too!