Hi,
I am currently in the process of designing MPLS network which will carry variant traffic types including voice related traffic (bearer, signaling, JTAPI, CTI, presence, ICCS, etc).
I am thinking to create CUCM cluster over WAN by placing servers in Dubai to cover EMEA region as primary (Dubai is the HQ of the comany) and servers in Kula Lampur to cover SE Asia/India region as primary.
The reason of creating single cluster instead of two clusters is to maintain features while users are roaming at any country office as well as simplify management (I am refering to features beyoned EMCC such as Jabber, Voice Mail, Presence, etc).
As per CUCM 9.x SRND, the requirements for WAN Clustering is to have max RTT latency of 80 msec. Also, the bandwdith per subscriber for BHCA < 10,000 should be N x 1.544 Mbps, where N is the total number of servers in other locations.
Using these values the bandwidth requirements will be high which is expensive in Dubai. Add to this, 80 msec is very restrictive value (the min RTT latency offered by ISPs between Dubai and KL is 220 msec).
I would like to hear from the experts their suggestions to move forward? Shall I take the risk and cluster over MPLS or do I need to start think about trunking two clusters which add more restrictions to features within a region