Added White Paper: Performance of CPS running on Citrix Provisioning Server - Wilco van Bragt - BrianMadden.com
Brian Madden Logo
Your independent source for desktop virtualization, consumerization, and enterprise mobility management.

Added White Paper: Performance of CPS running on Citrix Provisioning Server

Written on Nov 20 2007 2,983 views, 7 comments


by Wilco van Bragt

In a new article in the white paper section I'm describing the performance of Citrix Presentation Server running a streamed... Read More...

Read the complete post at http://sbc.vanbragt.net/mambo/updates/added-white-paper-performance-of-cps-running-on-citrix-provisioning-server.html

 
 





Comments

Guest wrote His test runs are bogus
on Tue, Nov 20 2007 5:08 PM Link To This Comment
The numbers don't added up in this article because he put the Provisioning Server (Ardence) on a Virtual Machine!  What a way to cripple and skew test results!
Guest wrote Re: His test runs are bogus
on Tue, Nov 20 2007 5:39 PM Link To This Comment

I am interested on why putting Provisioning Server on VmWare would cripple and skew the results?. There is no point in making a statment like above without explaining.

I was expecting slower performance for provisoning a server and with less concurrent users but as Wilco shows the results are pretty even and in some cases favours Citrix Provisioning Server .

Thanks Wilco I have been waitiing for this article for a while and is very good :)

 Dec

Guest wrote Re: His test runs are bogus
on Tue, Nov 20 2007 6:05 PM Link To This Comment

The question is "why would you choose to run Provisioning Server on VMware as your standard baseline for the test". It may run just fine, but you've written a paper specifically aimed at measuring before-after performance, then you've introduced an additional variable with no way of isolating what it's impact might be on the equation. I don't think Citrix has certified PVS running on VMW (or XenServer for that matter) so it's performance impact is unknown. Anyone who's worked with virtualization technologies will acknoledge that it can slow down the performance of some apps, so why introduce that question on a paper specifically aimed at answering a very useful question about PS and PVS performance. Would love to see the results replicated with PVS running native on the OS and see if there's any difference.

Wilco van Bragt wrote Why VMWare?
on Wed, Nov 21 2007 1:08 AM Link To This Comment

I personally don't think the tests are bogus, because we are only running one Ardence client connected to the Ardence Server. In practice I assume that you wants to connect more clients to the server. Beside that we choosed VMWare because the customer is using VMWare as the standard platform to build servers on. I agree that is would be perfect to run another test with a physical Provisioning Server and connect more clients to it, for more indepth performance details. However this test gives a overview of the performance and even with limited hardware showing that the results are at least the same as running a physical server.

Guest wrote Re: Why VMWare?
on Wed, Nov 21 2007 10:15 AM Link To This Comment

It's bogus because the Article is performance of CPS on PVS vs CPS on VMware.  If Vmware is that utilized at your customer, then the CPS test should have been done on ESX and NOT raw hardware.  And if the customer insist on use ESX for PVS, then perhaps that scenario is not best to publish to the rest of the world.

One connection vs multiple connection has no relevance on VMware or not when gathering statistical data.  You've introduced too many variables without isolating one of them.  That's why the numbers are bogus.  It's simply the scientific method.

 

Guest wrote Re: Re: Why VMWare?
on Thu, Nov 22 2007 4:12 AM Link To This Comment

I think people are mis reading what Wilco said or I am :)

Provisoning Server the Software is on VM. CPS was delivered to a pysichal server using Provisioning Server. He then had another physcial server with CPS installed.

 

 

Guest wrote Re: Re: Re: Why VMWare?
on Thu, Nov 22 2007 4:57 AM Link To This Comment

Wilco,

 I too agree with what the above poster has said. People are mis-interpreting your White Paper. They think that the Ardence client is on VMWare. Folks, this is how it is..... (Wilco , correct me if I am wrong)

 The Ardence Streaming server is a guest OS on the VMWARE ESX server (along with other guest OSes:Win2k3 AD,Win2k3 File Server, MS SQL 200. Look at Wilco's diagram half way down the page

The Physical servers are a Presentation Server ( traditional installation) and a Presentation Server (an Ardence Client)

The conclusion of your article is that the Presentation Server (Ardence Client) runs at the the same performance sometimes better than the other Physical Presentation Server...... What would it run like if the Ardence Streaming server were not a guest OS on the VMWare Server?....

 

 

Wilco van Bragt wrote Ardence Server is the only one on VMWare
on Fri, Nov 23 2007 1:43 AM Link To This Comment

The last two comments are completly right. Only the Ardence Server is running on VMWare. The CPS system were running on physical hardware. My opinion is that I mentioned that cleary in the article and also added an image of the set-up.

There is  possibilty if we configure the Ardence server that the performance of the Ardence Client even will be slightly better. But I must say that we did not see excessive resource usage on the ESX server during the test. Hopefully all people interpreting the white aper correctly now.

Guest wrote Where to find the white paper
on Fri, Jun 20 2008 3:24 PM Link To This Comment

Hi every oneIt seems like the homapge where the White Papier is published is down and has been it for some days now.Are there anyone who knows where I can find the White Paper today ?Thank you !Søren Bay /Topnordic Denmark

(Note: You must be logged in to post a comment.)

If you log in and nothing happens, delete your cookies from BrianMadden.com and try again. Sorry about that, but we had to make a one-time change to the cookie path when we migrated web servers.