We have an old Cisco CallManager 4.1(3) cluster (pub and sub) that we use to run a multi-customer VOIP service. We have a problem where one customer (Cisco phones are a T1 hop away from CM and PRI gateways) has echo during MeetMe conferences. All parties hear it, including those at the customer site and those who are on the PSTN and who talk through a PRI gateway, then through a 1 Gbps switch network to CM. This is the only situation where there has been an echo problem for a LOOOONG time, probably 4 years. I don't know right now how many participants are in the MeetMe conference. All transcoding is done in CallManager. Win2000 Pub and Sub have 2 GB of RAM, % RAM util runs at a steady 45%. CPU, disk I/O. memory paging all look very healthy all the time.
The same customer site has no echo or voice quality problems during normal calls, even when their T1 connecting to the data center where CM and gateways are is pegged at 100% utilization.
Is that version of CallManager not good at echo-cancellation? Should I be using hardware on the network for handling the transcoding of MeetMe conferences?