Jump to content
Welcome to our new Citrix community!
  • 0

GoToMeeting on Citrix CVAD 1912 LTSR


Marc Kuhn

Question

Hi guys

we have a Citrix CVAD environment with Server 2016 and CVAD 1912 LTSR. Today we tested GoToMeeting on the environment with 3 users logged in from external via Netscaler. The CPU Usage is going from 25 to 90%, depending on Webcam enabled or Screensharing. Also the Bandwidth is very high, so 3 users can easaly block a whole VDA without doing to much.

 

Is anyone using GoToMeeting in Citrix and is there any optimization we can do like with Teams?

 

Many thanks for your feedback


Best regards

Marc

Link to comment

10 answers to this question

Recommended Posts

  • 0

We use mainly Teams on Citrix as the new default to go platform (company policy decision) but I also see alot of Zoom happening nowadays as well as the occasional gotomeeting and equivalents. On the new platform anything works solid and I've heard nothing but good comments from end users while on the old platform (Win 2012r2 image with weaker HP G7 hardware and no GPU at all) I see nothing but problems nowadays. As such anything audio/video we don't even officially support on it within the company since playing a simple youtube HD video nowadays just flatlines all virtual cpu cores in the vda server.

 

For our new platform I did a Proof of Concept last year after alot of research and theorizing which eventually became our new production platform in the end. You can read up on my entire journey over at the Nvidia forums in this thread:

 

https://gridforums.nvidia.com/default/topic/10042/xenapp/looking-for-advice-on-optimal-config-for-latest-gen-citrix-xenapp-vgpu-solution/

 

(I'm profundido)

 

I've tested thoroughly with all kinds of audio/video content and this new platform just manhandles everything I throw at it so far. totally loving it already.

 

Hope you enjoy and if you would like me to perform a specific test for you on it, feel free to ask here or in pm.

 

  • Like 1
Link to comment
  • 0

Are you using Nvidia Grid cards to offload that insane cpu load as well as the reduce bandwidth ? Because that's exactly why that kind of solution exists. For my company's new 2020 platform I built new servers using HP DL380 G10 with Nvidia T4 cards and a brand new Windows Server 2019 golden image from scratch. The Citrix VDA servers encode that stream, send it over to the client using less bandwidth where it is decoded. In addition to that the VDA servers cpu's are also immensely offloaded for any audio/video related content.

 

Imho, this mechanic will become more and more necessary in the future as load keeps increasing.

Link to comment
  • 0

Hi Ftruyens1

 

no we do not use that kind of cards, but we are using also HP Server, normally we have Synergy Blades which we have configured. In that case we have couple of G9 blades configured. Many thanks for your input, i will check if we are able to install that kind of Cards and see if that is helping.

 

Are you using GoToMeeting or do you have any good expirience on Citrix except MS Teams? 

Link to comment
  • 0

Hi Ftruyens1

 

sorry, just a short question. I heard that we have a partner company, which has a server with a GPU, where we can do some test's. Is it true, that you can't install just Windows Server OS on this server? You will need to install a Virtualization like ESX or XenServer and just then are able to use that Card?

 

Many thanks for your feedback

 

Best regards,
Marc

Link to comment
  • 0

Hey, 

 

the key to solving any typical audio/video bottlenecks on the cpu is making sure the VDA server can offload to a any working GPU (even a consumer geforce/quadro/... card or whatever usually accelerates a local windows pc) in it. So purely for testing you could just simply modify and run your existing golden image config on any bare metal server (or install one from scratch) and then have people log into the Citrix VDA on that specific server. The Citrix and windows RDS policies you then configure to ensure offloading (hardware acceleration) happens towards the client machines etc. For instance in case your existing blades happen to run a bare metal install of Windows OS + Citrix VDA layer you could simply pop a GPU in there and test immediately. I didn't ask you if they were virtualized or not so I don't know.

 

 

But since in reality most production environments are nowadays virtualized on something like VMware where 1 host server hosts several VDA Citrix servers that your end users log in to, those host servers won't benefit from any classic GPU since it won't be usable in the virtual Citrix servers that run on the hypervisor OS. And that's where Nvidia grid comes into play. It allows you to put 1 or more specific (supported) GPU's into the physical host server and then use the Grid software to split them into multiple virtual software gpu's that you then can assign each to a different virtual Citrix server, pretty much like a physical GPU card becomes usable in a bare metal installation.

 

I hope this clarifies.

Link to comment
  • 0

Hey

we are using Hardware HP Blades as VDA's directly, most of our customers have physical VDA's. Are you using your HP DL's directly as VDA or do you have any Virtualization running on it and have configured virtual VDA's on that virtualization Layer?

 

You wrote this GPU is also reducing the bandwidth, is that correct?

Link to comment
  • 0

That's right. The combination of the right hardware, software version of Citrix and its policies allow the server to encode streaming data before sending it to the client (thus compressing it so less data must be sent over the line) rather than send it at its full bandwidth. The Citrix policies allow you to play around with that setting and tweak it. 

 

We are using Vmware virtualization with Nvidia grid software on the HP DL380's but you could try putting recent gpu cards into your bare metal blades to achieve a similar goal. So it will be very easy to test in your case.

Link to comment
  • 0

Sorry, perhaps i'm a little confused about what i was told internally. You are telling me it is possible to just put a GPU into our HP Blade or HP DL with Windows installed directly and can use the GPU? It isn't mandatory to have VMWare in between? That would be indeed an easy way to test.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...