TechReady5
Back from TechReady... Great week with some really interesting stuff. One of my personal favourites was a session on 'The Green DataCentre'. It's clear that Microsoft are doing a lot to be green but are definitely nowhere near where they should be in my opinion. Some of the other interesting stuff was;
DPM (Data Protection Manager) - loads of information on this about and definitely should be a very popular product with Exchange administrators. I really like the idea that for the future the responsibility for protecting and recovering Exchange will be in the hands of the Exchange administrator which is where it should have been all along. Have written about this previously.
SCR in SP1 - love the ReplayLagTime and TruncationLagTime features which give the administrator the ability to control the delay between a log being verified and copied to the target and it being played into the database. It is possible to have multiple targets with multiple ReplayLagTime's which might be interesting to many customers.
The TruncationLagTime can be used to place additional delay between when the transaction log is played in and when it is truncated or deleted. I think this will be used to provide an additional recovery option. So if a site failure causes a logical corruption which makes its way into the SCR database you now have the option of recovering from taped backup, brought back from off site, and bringing this up to date using all the logs prior to the corruption...
Windows 2008 - the cluster stuff is great - the GUI is miles better and the wizards are much more informative and easier to use. There is now loads of inbuilt verification to ensure that clusters are build out according to Microsoft best practises - along the same lines as ExBPA.
Only issue is the upgrade path from an existing Windows cluster. In all previous versions you can have multiple nodes of a cluster on different versions of Windows to ease the upgrade path. Problem in Windows 2008 is that this is not supported because clustering has been changed so dramatically. Only option for Exchange is therefore to build a new server and use database portability or move-mailbox. Not ideal if space is at a premium but with most customers moving to 64 bit it might not be as painful as it seems as a lot of companies might take this opportunity to replace their hardware. Of course you don't have to completely build an entirely new cluster - you could just evict one node at a time and rebuild each one into a new cluster. This represents a perfectly feasible upgrade path for the O\S - it's just the applications that complicate this approach.
Also went to a few OCS sessions - it may take a few years to be widely adopted but E2K7 with OCS and IP telephony is going to be massive in my opinion...