THE BriForum Session 10


So this year for BriForum I have submitted two sessions so far and as one of these is potentially massive, here I am asking for help.

The session will basically attempt to cram in 75 minutes a review of the so called Citrix Best Practices and check which ones still hold water to this day. The main reason for that is I still see, believe me, people looking for the (in)famous “Metaframe Tuning Tips” that my buddy RickD wrote when dinosaurs roamed the earth. Seriously, the reasons are many. To this day I still see many projects being done following these best practices and I am certain, given how hardware, software and solutions in this space have been evolving, many do not apply or have to be slightly modified to still be valid today.

Now the question you are probably asking yourself now is, “Well you are the Mr. CTP/MVP so you know better what is valid or not, what these best practices are and so on. So why the hell are you asking us?”. Well first of all my time is limited. That means I cannot be working on 1,000 projects at the same time. But if I ask 1,000 people in the community that are currently working on projects I am sure we will be able to gather a lot of great info that I would not be able to get and/or see myself.

And finally I do not know it all. Well no one does. Not even Shawn Bass. The power of this community I think is what can make a session like this great.

All that said I would like to ask you guys a couple things:

– What do you see to this day in terms of so called best practices? As an example I see all the time “do not use virtual PVS servers”. So what are these best practices?
– Any best practices you know for sure should be discarded and never used again? Example: some NT4 registry settings I am sure do not do shit on 2008 R2 or they actually do shit on your environment, throwing shit at the fan for the fun of it.
– If you will be at BriForum, are you brave enough to join me and a bunch of other clowns on stage? As you know I am probably one of the most well behaved, polite and politically correct presenters at BriForum, not to mention the All-Time GeekOut champion. So fear nothing, you will be well treated and gain respect from your peers (not me. LOL).

Resuming: whatever you have to say in terms of best practices (good, bad, old, new, etc) PLEASE let’s talk. Feel free to either post here or to email me directly at CR at WTSLabs dot com. If you put [BriForum Session] in the subject, that would help a lot sorting out the two emails I am expecting.

Cheers guys, appreciated. See you (hopefully) at BriForum.

CR

108,383 total views, 2 views today


Leave a comment

Your email address will not be published. Required fields are marked *

10 thoughts on “THE BriForum Session

  • Jennifer Auiler

    I will most certainly have some material for you on this. I just got done with a reasonably sized XA6.5/PVS deployment and wow did I learn more about Citrix than I had ever imagined I would. How quickly do you need this?

    • crod Post author

      Well the sooner the better Jennifer! Not in a major rush as BriForum is in May but I do expect a lot of content to cover so I am trying to gather as much as I can by April.

  • Jarian Gibson

    You can virtualize PVS. Just make to review infrastructure and setup properly. Same goes for CIFS PVS stores, just make sure to setup properly with SMB tuning etc on Windows Server 2008 R2. I tend to stay with vDisks on local PVS storage. If you have physical PVS servers, why put vDisks on any other place than local disks? Make sure to size PVS RAM correctly for number of vDisk images being streamed. Again with correctly RAM sizing, it is ok to use dynamic vDisks with PVS. Don’t use private mode vDisks as dedicated images. Use MCS, MCS with PvD, or PVS with PvD for this. EdgeSight agent can cause performance overhead especially when EdgeSight agent logs are placed on the PVS write cache disk. Use fixed disks for write cache disks. EdgeSight with MCS need to use Agent Database Broker. NFS is king with MCS.

  • Bryan Mann

    Agree with Jarian, it can be a VM with some caveats. Be sure that the PVS server vNIC is a seperate HARDWARE NIC from the targets if it is on the same host. Avoids queuing/buffering. Would love to see pagefile Best Practice addressed. We have had good success turning of the pagefile in our XD VMs and XA VMs when using PVS. Cuts down on storage needs for the WBC.

  • Dan Fallon

    I agree on virtualizing PVS, and I like Jarian’s suggestion of local storage on PVS server for vDisks, even with virtual PVS servers. I still dont trust CIFS 100% because of the required registry tweaks, even though I know CCS has verified them.
    Im curious if Bryan’s recommendation to give virtual PVS server a dedicated NIC was for a 1G environment? I’ve done 10G environments where all VLANs were trunked over redundant 10G, no specific h/w dedicated to PVS.

    • crod Post author

      I personally did both virtual PVSs and CIFS shares (with the proper tweaks, etc). I can attest they do work properly, if correctly sized (file server RAM is also critical). I can see this session has a lot of potential for BriForum!

    • Bryan Mann

      Dan,
      The situation I am referring to is if you would have a PVS VM on the same HOST as the XenApp/Desktop VMS. If they use the same nic, we have seen, at least with XS, that the packets, buffer and queue because it never leaves the NIC so to say. As soon as we put the PVS VM on a different NIC from the XenApp VMS the issue went away and packets were sent immediately.

      • Dan Fallon

        Interesting, so it was an actual functional issue where PVS streaming didn’t work until PVS was on a different server or was it performance issue? Have never seen that on vSphere, but normally have a dedicated cluster for infrastructure severs like PVS, although that isn’t always possible in small environments.