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

Hypervisor vs. Citrix Virtual Apps and Desktops Service


Question

Can someone explain what the difference is between the two products for the following scenario:

 

1 Host

4 Server VMs (Server 2019)

85 users connecting into shared server connections (i.e. approx. 40 users per VM simultaneously)

Each VM would run apps specific to our work along with O365

 

I wasnt planning on running 85 Win10 desktops in order to keep costs down. Shared sessions made more sense in this case. My thought was to use Hypervisor, build the VMs and have users connect to the VMs. I wanted to do all of this via Citrix instead of RDS since there will be GPU involved in some cases. Why would I need the Citrix Virtual Apps and Desktops Service?

Link to comment

11 answers to this question

Recommended Posts

  • 1

You could do that, yes. Hyper-V running Windows Server 2019 was very stable from my experience, but XenServer and running Windows Server VMs on it will work, as well. You'll need XenServer licensing if you don't also run Citrix Virtual Apps/Desktops and want the supported paid-for XenServer edition features, which I believe you'll need since you want GPU/vGPU support.  If you get Citrix Virtual APps/Desktop licenses, you get all the features supported in the then free XenServer/Citrix Hypervisor platform as a bonus. The nice thing about that setup would be the ability to chop an existing GPU into vGPUs and allocate to your VMs as needed (plus you can of course also support multiple GPUs on a single host, if desired). But then you'd have to run it as a Virtual Desktop/Apps server and not RDSH (assuming I fully understood your plan!).

 

-=Tobias

  • Like 1
Link to comment
  • 0

You have a number of options here. You could run Windows Server (one or more instances) as a VM and allow multiple users to connect to Windows Server as a VM and share a GPU assigned to the VM. You could also run individual Windows 10 desktops and assign a vGPU to each. You can also use RDS and run on bare metal and assign a GPU to a Windows server or using DDA, split multiple GPUs on the same server between VMs.

 

A lot will depend on factors like Microsoft licensing, GPU licensing (where applicable), and if you want or need to guarantee a certain amount of GPU capacity to each VM or user. Both approaches -- XenServer and Windows Server VMs with multiple concurrent users as well as a pure Hyper-V instance with multiple Windows Server VMs and GPUs assigned via DDA will work.  If you go with NVIDIA GRID GPUs, the licensing gets more complicated because you have to factor in server licenses as well as user licenses.

 

Also, 40 or so users per VM may be a lot depending on what tasks they are dunning and various other resources (CPU, memory, storage IO, network bandwidth, etc.). The entire environment has to be designed so that there isn't a major bottleneck anywhere.

 

This is the sort of situation where a consultant can help you make good decisions based on your environment and needs as there re a number of good options, but where also details really matter.

 

-=Tobias

Link to comment
  • 0

You could do either. The main difference, if I understand your model correctly, would be using Windows servers under Hyper-V vs. Windows servers under Citrix Apps (and presumably Citrix Hyperevisor as the hypervisor platform). Where I worked we ran both, the Citrix Apps mainly because of some vendor license requirements. Both setups had multiple GPUs hosted per server. In both cases, you need the Microsoft licensing and could save the extra Citrix licensing by staying with a pure Microsoft model. As mentioned before, DDA can provide the means under Hyper-V to hosts multiple GPUs per server.

 

-=Tobias

Link to comment
  • 0

Tobias -- thanks for the additional information. Its been a while since I have used Citrix (6 years) so maybe I am overlooking something. Can I install the 4 VMs on Citrix Hypervisor? I'd prefer not to use Hyper-V since vGPU isn't as stable (at least this is what I have been told). My ideas was to install Citrix Hypervisor, load the 4 VMs onto it and then allow for RDSH on those 4 VMs. Am I correct in thinking that this will work?

Link to comment
  • 0

Each VM was assigned a full P4 (8Q NVIDIA profile). You have to think this s an educational environment, so not heavy-duty users like in industry. It never saturated, hence I'd guess 20 - 30 concurrent users would not be overloading it. Much of the time with 10 - 15 users it ws just using 10-20% of the GPU cycles. Each VM had 16 VCPUs and around 100 GB of RAM assigned to it.

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...