I’m not really an Exchange person. In my jobs, I’ve either worked in places with Lotus Notes (and dedicated mail admins) or in environments where Exchange was tiny.
I’ve been attending Nathan Winters’ sessions on Exchange 2010 today in Microsoft Ireland. Of interest:
- The Outlook client sends all communications to Exchange via the CAS. It doesn’t directly talk to the Mailbox servers. Exception is for public folders.
- DAG: Exchange mailbox servers replicate data around the network (LAN or WAN) for fault tolerance. It’s not Windows Clustering (that’s gone from Exchange now). Up to 16 copies of your data.
- Built in basic archiving. Don’t use PST’s stored on file servers because (a) you’re just moving the problem and (b) you’re wasting expensive primary disk.
- New storage model optimised for SATA disk. This bigger disk approach uses streams. Therefore no more single instance storage. No need to use a SAN. JBOD might be the best approach. Using these bigger and cheaper disks will make Exchange 2010 more scalable and cheaper. It’s a requirement for unified communications.
- No more single copy clusters, i.e. tha cluster with shared disk – a single point of failure. Use JBOD and DAG for fault tolerance. Probably cheaper anyway, e.g. I can get 35TB of HP JBOD for €21,000 (recently priced) VS €13,000 for a basic HP cluster kit with next to no disk in it (priced 1 year ago).
- Easier federation via Microsoft online service for integrating different Exchange organisations, e.g. company merger or inter-company partnerships.
- LOADS of new end user functionality.