OK, so another day of Exchangeness has been and gone, here are my highlights.
The keynote this morning was “Geek out with Perry!” with Microsoft’s Perry Clarke.
The format of this was intended to start the “interactive” element of MEC with Perry responding to (scripted) questions from members of the audience, unfortunately it was a little slow and lacking energy, especial as it was the first session after a night at Universal Islands of Adventure with all the food & drink you could ask for!
The presentation was performed on an 87″ Perceptive Pixel (recently purchased by Microsoft) touch screen where Perry was using OneNote to illustrate his answers.
Topics covered were
- The use of “cheap” drives rather than “enterprise disk” (only a 5% difference in failure rate)
- SSD (Not a good fit for Exchange from usage patterns or ROI perspective)
- Will MS’ focus on the cloud reduce the emphasis of on-premise products (No!, experience from cloud service will add features and functions to on-prem)
- Lowering of IOPS (99% over the last 10 years)
And finally, what is the upgrade story from Exchange 2003 to Exchange 2013. “It’s at least as good as the 5.5 to 2007 story!”. “These organisations are obviously risk averse as they are still running 13 year old technology”.
- Move to the cloud, there will not need to be any ROI justification for future upgrades
- Move to Exchange 2010 now, then migrate to Exchange 2013 SP1 when you are ready.
Exchange 2013 High Availability
This session was interactive (a unique feature of MEC, encouraging 2-way communication between the experts and delegates), so discussion were a little “off-topic” at times.
The Experts leading the session were Greg Thiel & Scott Schnoll, both very experienced and knowledgeable with a good sense of humour.
Key information from the session
- Automatic site failover requires the Witness server to be on a separate, 3rd, site.
- With only 2 sites available site failover performs like 2010 BUT can fail roles only if wanted (e.g. just CAS role).
- If a DAG node drops out of a cluster for 5 minutes, but can still be contacted via RPC then it won’t be failed over.
- The new Auto re-seed capability in 2013 DAG’s requires the use of volume mount points.
- Windows Server 2012 Storage Spaces will be supported in 2013, Deduplication won’t.
- Datacentre failover will prompt a client redirect (HTTP) using Geo-DNS, this may not work on some ‘phones as their redirect behaviour varies from device to device.
- If a server is put into Maintenance mode by SCOM it won’t affect Managed Availability.
An interesting conversation occurred around DR sites, Greg suggesting the DR site DAG member(s) should be used actively all the time in production as this proves the site & service rather than hoping it will all work in a disaster.
Exchange 2013 Database and Store Enhancements
A rather heavy interactive session full of stats and specs
- 50% IOPS reduction from 2010 to 2013.
- By supporting multiple databases per volume IOPS are maximised.
- Isolation of individual store issues is achieved by having a store process per database.
- A single, badly behaving, mailbox cannot impact the entire database.
- Database Schema elements are optimised for Sequential I/O
- All Exchange virtual machines should use Fixed RAM not dynamic.
Exchange 2013 Upgrade for Exchange 2007 and Exchange 2010
A very interesting session, for more than one reason!
Some key points:-
- Co-existence will require 2010 SP3 or a 2007 rollup, both to be available early next year
- Outlook 2003 is not supported with Exchange 2013
- New & old Public folders cannot co-exist (I may need to get further clarification on this)
- Active Directory will need a 2008R2 DC but only needs 2003 Functional Level
Pretty much for any protocol the process is to move the namespace to point to the Exchange 2013 CAS and that will deal with proxying requests to the legacy servers.
One less appealing fact was highlighted – OWA redirecting to 2007 via a 2013 CAS will prompt a second time for authentication – not ideal. This may be resolved in the released product.
An interesting discussion then broke out around the discontinuation of TMG, over 75% of delegates in the room raised their hands when asked who used TMG with Exchange currently. The Microsoft line is “you have options”, stick with TMG, use UAG, have no Reverse Proxy. The current Exchange 2010 rules in TMG will need slight modifications to support 2013.
Oh, and I got to play with the 87″ touch screen in the exhibition hall – I need one of those!
Final day tomorrow and the trip home, watch this space for my updates