I recently upgraded from CUP 7.0 to Presence 8.0 (with CUPC 8). I have 2 nodes in a single cluster. In presence 7.0, there is automatic failover to the backup node when the primary node is down.
Now I am very surprised to discover in the release notes of CUP8.0 that the High Availability is no more supported!
"If you deploy the active/standby user assignment model in Cisco Unified Presence Release 7.0(x), and you upgrade to Cisco Unified Presence Release 8.0(x), where the active/standby model no longer supports failover, we recommend that you redistribute (rebalance) your users across the cluster. "
(Extract from Deployment Guide 8.0)
It is a shame to release a major version losing an important feature like that! How can we explain to customers that they bought a cluster to have redundancy but they don't?
Does anyone got the information about the roadmap when this feature will be reimplemented?
PS: CUPC8+CSF takes >200MB of the workstations RAM! It's huge!
First of all, you should ask yourself "Why do I (or my clients) want to upgrade to CUPS 8 if CUPS 7 was working perfectly?".
Cisco build different software "trains" in parallel. For example: they build CUCM 7 and CUCM 8 in parallel. So it's pretty normal that CUCM 7.1.5 comes out after CUCM 8.0.1. Just like Exchange 2007 SP3 comes out after Exchange 2010 was released.
Which version is better? Shouldn't the "latest" be the "greatest"? It depends.
Usually, a new "major version" was released because the architecture was changed dramatically (because the requirement of new features). Since different' "major versions" were being developed in parallel, it's very difficult (sometimes impossible) to synchronize all the features and bug fixes to each other. For a customer that was satisfied with the features of 7.x, version 7.1.5 would be the best version. For customers that would like the try the features on 8.x, they might have to bear the bugs in 8.0.1 (which might have already been fixed in 7.1.5).
Back to CUPS. The "new feature" of 8.0 is the incorporation of "Jabber" (XMPP). XMPP is supposed to be the next standard protocol for presence and instant message and being used by many big names such as Google, Microsoft, Yahoo, etc. Since Jabber was acquired by Cisco, it takes time to "merge" into CUPS. That's why the first release didn't have HA. More specifically, didn't have HA on XMPP. HA support is scheduled in CUPS 8.5, which is due the end of this year.
Someone might argue, that Cisco could have waited for HA before release the 8.x version. Well, that's debatable, because some customers out there want XMPP so badly that they don't care HA at all.
Same for the CSF architecture. It's currently developled with Java, so memory and CPU usage won't be optimum (well-known issue of Java). Cisco is building C++ version which would significantly improve both memory and CPU.