CCR may be the most interesting apsect of Exchange 2007 that I’ve come across so far. The one aspect of MS solutions that I’ve had a big probelm with is their implementation (or lack thereof) of a a true Disaster Recovery solutoin, i.e. a geo-cluster. What’s a geo-cluster? Imagine your office burns down and you have a DR site. If you are a big financial or government then you may only have X hours to invoke and get the business back up and running. MS clustering for SQL 2000 and Exchange 2000/2003 just weren’t up to this. Lotus Notes *choke* freaks loved to remind us that they had a true DR clustering solution and that Exchange didn’t. They were right. I hate saying that.
But the worm may have turned! Exchange 2007 includes a function called Cluster Continuous Replication. The concept is that the mailbox server can be configured to replicate mailbox databases to a "passive" node. In the event of a DR invocation, you can failover and use the formerly "passive" node as your "active" node. Here’s where things start to get nice. You don’t need a shared SAN. The two nodes can use their own locally attached disk.
I’ve not dug into CCR too much yet. I’ve only speed read the above article. This may be a solution that will finally allow us to rid ourselves of EMC Autostart, Neverfail and their ilk for Exchange DR.