Sample LTM Configuration for OCS 2007 Enterprise Pool with Scaled-site Edge servers Deployment

Thanks - I used similar MS documentation to do my initial config (we are not using OCS R2, so it's a little different). MS talks about "TCP-level affinity" and F5 allows you to configure "persistence" - I'm not totally convinced they are the same thing, but if they are, which type of persistence is required?

Hello

 

 

Anything new about these details?

 

 

Issue is with OCS A/V Edge (running in cluster) and NAT device (F5 LTM).

 

 

WBR "Woodpecker"

Hi WBR - I disabled persistence for all my virtual servers and that seemed to fix the issue of not being able to use both of my AV edge servers at the same time. (Thanks to a member of this forum responding to another one of my posts.)

 

 

We also opened a trouble ticket with F5 support to have them validate our configuration. That process is almost complete (they are currently reviewing a TCPDump), and so far there is nothing really wrong with our configuration. I do have a latency issue with some of my users' calls, but I think that is due to network conditions at some of my user sites.

Empty

Hi Joe,

 

 

did you get the validation of the configuration from F5?

 

 

I'm hitting the following problem: when a front end dies, users correctly register on the other font end, but are not able to rejoin the A/V conference that was "hosted" by the died front end. From my investigation I suspect this could be related to the persistence configuration on the F5. So it would be interesting to have an official word on how this should be configured for each Virtual Server.

 

 

Regards

 

Gianluca

Hi

 

 

I talked to MS about this load balancer and persistence issues and finally I asked them if they (meaning MS) wanted destination or source persistence. They said source persistence.

 

 

We've setup the F5 with source address persistence and now when a front end dies, the users are are sent to the other front end server. There still the issue where the conference originator cannot rejoin the conference unless some other attendee drags and drops him in, but at least there is failover.

I eventually turned off persistence completely for all my VS's. This solved my immediate issue of being able to use two edge servers. However, I did not test the capability to kill a front-end during a conference and rejoin that conference. I did have F5 thoroughly review my configuration and they didn't see anything blatently wrong with it.

 

 

- Joe

Hi all,

 

 

I have also tried to disable all the persistence setting in the virtual server, but i 've also encountered problems when using the web conferencing application between the internal users and external internet OCS client users. I am currently using BIGIP 10.1.0,are there solution for this case?