We currently run a pair of (about to be) clustered VCS-Controls that all our endpoints register to - we'll call these EndpointVCS1 and Endpoint VCS2. Our endpoints can then use DNS SRV to lookup SIP and H323 details for ourcompany.com, giving details for EndpointVCS1 and EndpointVCS2 with different weight/priority etc.
We also run a second set of VCS-Controls that our Telepresence Servers register to - we'll call these MCU-VCS1 and MCU-VCS2. Currently they are not clustered. As we can only have a single SRV entry for our domain, Currently, MCU1 can only use MCU-VCS1 and MCU2 can only use MCU-VCS2, as we can't make seperate SRV records for the MCU-VCS's.
So my quesiton is this - how can I achieve redundancy with two seperate VCS clusters in the same domain? Are there any plans to improve clustering and redundancy for VCS/Telepresence server, such as the way CUCM/CTMS works where you can enter multiple hosts?
When reading your post the first question that comes to mind is: why do you have four separate VCS installations? Have you really grown beyond the capacity of a single cluster; or is there some other architectual reason why you're forced to retain both Endpoint-VCS1 and -VCS2? The second question is why are your TPS registered to separate VCS Control servers?
The traditional way VCS was designed when you had multiple clusters was to use subdomains. For example, if the clusters serve different continents they could become na.domain.com and emea.domain.com. VCS follows the DNS heirarchy-based naming principals fairly well. There are methods of using FindMe accounts on a VCS Expressway cluster to front-end multiple Control clusters for single "public" identities; however, this is extremely manual since you have to configure it manually per-user (e.g. firstname.lastname@example.org is a FindMe account with email@example.com as a device).
My opinion is that you should not expect the basic design concepts of VCS itself to change. Without getting into product roadmaps - which require NDA signature - Cisco UC Manager is likely the way to solve this when URI Dialing eventually gets added. Yes, that's a tightrope you see me walking.
Thanks. No we are not exceeding capacity of a single cluster, it's for architechual reasons.
Our "Endpoint VCS's" are for our internal use only. The "MCU VCS's" are managed by us, but other organisations share our MCU resources via VCS Expressway..
Would it be viable to set up a seperate, dedicated DNS for the MCU VCS's and Telepresence Servers? Would it matter if they were managed by TMS if TMS and VCS used different VCS's?
You might want to investigate new search feature plan to be supported in next software release (X7.2).
Search rule will allow configuring “subzone” level in calling source. (Currently only support up to zone level.).
So you may create two subzones in VCS, one for internal Endpoint and other for shared infrastructure.
You able to secure convertibility by configuring search rule where call initiated from and destination zone/subzone.
This way without adding directly Gatekeeper to handle multiple cluster VCSs with single DNS SRV, you may utilize all VCS resources better way.
That sounds like quite a useful feature. For my particular problem I don't think it will help as the reasons for multiple VCS clusters are partially political and I can't change it unfortunately!