Cisco Support Community
Community Member

Voice media stream application deployment within a ccm cluster


I have a CCM cluster of two machines: The publisher and a subscriber. The subscriber gets the bulk of devices registered to it due to the power of the platform (7845).

I do not know where to start the Voice media stream application because the CallManager System Guide states ' The Cisco IP Voice Media Streaming Application service may reside on more than one server in a Cisco CallManager cluster, but it must not be the publisher server or the server that is used for the Cisco CallManager service.'

Where could i start it ? Because according with the documentation, my publisher is not the right place, and the subscriber, which is being used for call processing is also not recommended .


Re: Voice media stream application deployment within a ccm clust

There are really two questions:

1) On which server should you be running IP Voice Media Streaming Server?

Both. You have two servers, and the publisher needs to be a full and complete backup for the subscriber. Which one ends up being (the most) busy is determined by your media resource groups and group lists, which brings us to:

2) Which server should handle the primary streaming duty?

This is going to be a judgement call. If your publisher and subscriber platform differ substantially and you are mixing up your phone registration load unevenly, there will be no documented best common practice that will apply cleanly to your situation.

If you define your media resource groups and group lists such that all phones use one server, the other server's voice media streaming application will be completely idle (with the exception of any multicast MOH sources configured for continuous streaming). It will be fully available for service should your primary streaming server fail. This, you definitely want to do.

How to split the load between your publisher and subscriber is, again, a judgement call. Usually it's a 100% one-or-the-other thing on a two-server setup, but yours is non-standard. You should consider measuring CPU and memory load on each server and distribute load around using MRG/MRGLs on phones or device pools as appropriate to your situation.

I don't know of any publicly available documentation for unbalanced publisher/subscriber deployments, so this is my personal opinion only. If you still have questions, and especially if you have a very large and complex deployment, please approach TAC or your local Cisco account team for guidance.

CreatePlease to create content