We have been running BES 3.6, so I am not sure how valuable this information will be to you, however...
We have 2 Exchanges servers, each in a different physical locations.
We have one BES 3.6 SP2 in one of the two locations serving both Exchange servers
We have around 120 BES users, with about 60 users from each of the Exchange servers.
I am not sure exactly when it started, but at some point users on the remote Exchange server began seeing message deliverly delays (from 5 - 2 hours). We begain receiving lots of "Some worker threads have been blocked for 31 health checks" or "At least one worker thread seems to be blocked (2)" warnings in the event log. I expect it was after we added additional users from the remote Exchange server.
According to RIM, this was due to the efficiency (or lack of) of the MAPI protocol and the way it expects to open and close its sessions during a specific time window. The latency between our two sites was 90-120 ms. Microsoft's recommend latency (and therefore RIM's) is 32 ms. RIM's recommendation was to deploy a BES to the remote site. In other words, the information the guys above had posted is dead on with what we saw.
We asked if BES 4.0 would correct the problem or inefficiency. They said it would help, by closing hung threads after a period of time and and not letting the server kill itself. But, it would not solve the overall problem. Never did they suggest just deploying the BlackBerry router to the remote site.
As for the multiple Exchange routing group, I don't think that really comes into play here. The key point is that the BES and Exchange server(s) are seperated by a WAN.
We have since deployed a BES 4.0 SP3 HF4 server to the remote site and have begun moving over users. As expected, users that have been moved are no longer seeing delays.
Last edited by Wakefield103 : 03-01-2006 at 02:45 PM.