Recently Apple released iOS 4 (the new name for the operating system that runs on iPhones, iPod touches, and iPads). Since its release there have been numerous reports (link, link, link) of a numbe...
Published Jul 01, 2010
Version 1.0
Blog Post
Our single Exchange 2003 server gradually over the course of about 4 days came to a grinding halt after many of our users upgraded to the iPhone 4.
Last Friday I noticed that the full backup of our 275GB mailbox database took almost twice as long to complete (16 hours instead of the normal 8 hours). The next couple nights were equally bad, and then the business day started on Monday and almost immediately the normal daily email was queuing up and messages were being delivered up to 90 min late. The server just couldn't keep up, but this didn't really make any sense because everything was working properly, but it was just not fast enough. What possibly would cause such a sudden and large performance hit? Eventually we tracked down that it was the iPhones, so we stopped it at the firewall, witnessed Exchange start to improve almost immediately, and then we let the mail queues catch up. At that point we decided to open the necessary ports once again, and things remained fine for a couple hours until... all of a sudden the mail queues jumped way up again. The mailbox store actually began to refuse everything. Outlook clients wouldn't send mail at all at that point and got immediate bouncebacks, so of course our users weren't too happy about this, and neither were we. The application log was littered with events from Exchange, and I was extremely concerned about possible database corruption. The store.exe process was constantly churning away, but no mail was coming or going, and fortunately a simple restart fixed everything. But that was only after a lot of sweating since the previous few night's backup had not completed yet either.
We have since gone back to using just IMAP for now, so that our Exchange server can keep up until we upgrade to a 4-server Exchange 2010 solution next month, but of course our users want the calendar and contact sync functionality from ActiveSync, as well as the push email. I would love to see Apple and Microsoft address this immediately.
Last Friday I noticed that the full backup of our 275GB mailbox database took almost twice as long to complete (16 hours instead of the normal 8 hours). The next couple nights were equally bad, and then the business day started on Monday and almost immediately the normal daily email was queuing up and messages were being delivered up to 90 min late. The server just couldn't keep up, but this didn't really make any sense because everything was working properly, but it was just not fast enough. What possibly would cause such a sudden and large performance hit? Eventually we tracked down that it was the iPhones, so we stopped it at the firewall, witnessed Exchange start to improve almost immediately, and then we let the mail queues catch up. At that point we decided to open the necessary ports once again, and things remained fine for a couple hours until... all of a sudden the mail queues jumped way up again. The mailbox store actually began to refuse everything. Outlook clients wouldn't send mail at all at that point and got immediate bouncebacks, so of course our users weren't too happy about this, and neither were we. The application log was littered with events from Exchange, and I was extremely concerned about possible database corruption. The store.exe process was constantly churning away, but no mail was coming or going, and fortunately a simple restart fixed everything. But that was only after a lot of sweating since the previous few night's backup had not completed yet either.
We have since gone back to using just IMAP for now, so that our Exchange server can keep up until we upgrade to a 4-server Exchange 2010 solution next month, but of course our users want the calendar and contact sync functionality from ActiveSync, as well as the push email. I would love to see Apple and Microsoft address this immediately.