Saturday, 4 August 2012

XenDesktop 5.6 with Windows NLB, 1000 VDI’s and Wyse T10 Terminals

Here’s a nice long question for any Citrix Gurus out there who happen to come across this post!


If I have a Citrix XenDesktop 5.6 infrastructure, presenting around 1000 desktops, using Provisioning Services to stream to Windows 7 virtual machines sitting on top of a VMware infrastructure, and delivered by three Citrix XenDesktop controllers, via two web interface servers using Windows NLB, to a site less than 50 miles away, using a 10 Gbps pipe; is there anything wrong with this configuration?

Fig. 1000 XenDesktop VDI Infrastructure (the mirrored SQL database is not included in the graphic)

One answer:

Windows NLB?

I would suggest using a couple of Citrix Netscalers to load balance connections to the Citrix Web Interface servers instead of NLB.
Readers, please let me know what you think!

Fig. Citrix Netscalers

Without going into too much detail:

Why am I asking this question?

If you had Windows embedded thin clients with Citrix Receiver, and they worked well with the XenDesktop infrastructure (the very rare odd grey screen); and then you bought Wyse T10 thin clients (which are excellent thin clients) with Wyse’s proprietary ThinOS, and then the T10s were having problems with random logouts to CTRL + ALT + DEL, random disconnects with “connection reset by peer”, and WTOS_log.txt was showing “ICA: Connection closed by server!”; the first thing you would think - with the Wyse T10s being the new components in the infrastructure - is that the problem is with the Wyse T10’s and their ThinOS, but ….


  1. I think using the netscalers is good, one step further would be to use the built-in web interface on the netscalers to keep the infrastucture even simpler.

  2. I remember in 2011 after I updated WI, similar problem start. So, reverced back. Could be same case