I am seeing the same thing. It started almost 3 weeks ago. I run a probe every 10 minutes to see response time to/from a blackberry device. In one 10 minute segment, I get between a 5 and 10 minute delay, on the other 10 minute cycle I get 3 - 8 seconds. It repeats like this all day. It appears to be doing a rescan every 20 minutes on everyone. On the short response time interval, I can send a <confirm> to any device and get a response immediately. During the long response time frame, I can send a <confirm> and not see the response for upwards of 10 minutes. RIM thinks the Microsoft patch 902400 is to blame, but we have had this patch since October. We have run sniffers on our network between the BES server and the Exchange boxes and nothing substantial stands out. The traffic looks the same during the delay timeframe and the immediate response timeframe. I am wondering just how many companies out there are actually getting these same issues and if they occured around the same time as ours. Also, has anyone applied any patches on their Exchange or Windows server environments prior to this problem occuring? We applied the MS07-026 patch the weekend before this started, though I can't see anything that changed in it that could affect us this way.
If anyone finds an answer to this, I would appreciate it if they passed it on. I am out of ideas for what to try to fix it and RIM seems to be stumped as well.