The second day at the Microsoft Exchange Conference (MEC) in Austin was possibly the best day of sessions and content that I have ever experienced at a Microsoft technical conference. That’s not a statement that I make easily, so let me explain my day and why I think it was so good.
First stop was a spirited and passionate debate about public folders and site mailboxes. Well, site mailboxes were dealt with in one question and the rest of the time were spent discussing the limitations that exist in modern public folders, why those limitations exist, and what Microsoft is doing to make sure that customers can actually migrate their existing public folder implementations to the new version. The session ended on a very positive note with the development group being given three clear priorities: fix the scalability issues, add back support for public folders used for shared calendars and contacts, and provide better management and monitoring tools.
I moved on to a great session on High Availability. By now everyone should be aware of the amount of effort that Microsoft has put into building high availability features into Exchange 2010 and 2013. What came out in this session is the continued effort to refine and improve high availability and some excellent and well-founded pieces of advice from eminence grise Greg Thiel about the need to simplify and automate. Nuggets like “humans are the biggest threat to recovery times.” Why? Because we all tend to make mistakes under pressure, which is why you should automate the response to problems. Managed Availability was covered too and an internal utility used to show how Microsoft monitors MA alerts in Office 365. It would be nice if the same utility was available to on-premises customers.
Next up was a discussion about “Clutter”, the new project to liberate user mailboxes from a lot of inbound mail that apparently shouldn’t claim any of our “mental energy.” Think of it as Inbox rules on steroids with a healthy dose of machine learning providing the smarts to identify on a user-by-user basis what’s important to an individual and what’s not. Jim Edelen told me that his mailbox was cleaned of some 177,000 unread messages when he ran clutter on it. The thought of so much unread mail made me sick but it’s good to have an answer for people like Jim who apparently can’t cope with the volume of inbound mail they receive. Clutter will show up for OWA users of Office 365 before the end of 2014 and I heard that it will also debut in mobile clients and in Outlook (including some legacy versions) as well, but not until sometime (maybe early) in 2015.
The session also covered People View and “search on lightening”, which is apparently needed because email searches are so much slower than web searches, something that I hadn’t noticed up to now. People View means that Exchange learns about the people who are important to you (by observing the people that you send messages to) and then constructs views that allow you to filter your inbox by these people – so you can click on “Tony” (for instance) to see all waiting messages from me. Again, these features will show up in OWA first in Office 365. Plans are still being worked on for on-premises release but I wouldn’t be surprised if they show up in the new version of Exchange expected toward the end of 2015.
I dropped into a superb session from Vivek Sharma telling people about how Microsoft runs Exchange Online within Office 365. The impression here is that standardization (in everything), automation, attention to detail, and the total commitment of the engineering team are strong elements behind the ability of Office 365 to deliver a service based on 100,000 servers. Lots of good detail about Microoft’s capacity planning and workflow systems. Best quote from this session was “we don’t let marketing touch servers.” Quite right too!
My last session was Jeff Mealiffe’s discussion about best practices for virtualization of Exchange. Now, my personal preference has always tended to be to have mailbox servers on nice physical servers and this impression was confirmed and deepened by Jeff’s remarks, which were practical and detailed. There are many pitfalls that await the unwary who venture into virtualized Exchange in production (test boxes are great on virtual machines) so it pays for anyone who wants to do this to prepare diligently and make sure that they can support the deployment. Speaking of support, we covered the whole “why Exchange doesn’t support NFS” topic again. In a nutshell, NFS is not supported and only 3 people in the crowd of several hundred were interested in NFS anyway, so it doesn’t appear to be as hot a topic as imagined.
In any case, MEC day 2 was great. I enjoyed all the sessions I attended and believe that the same opinion is shared by most other attendees. And now on to day 3 when I’m involved in two sessions. Should be a blast!
Follow Tony @12Knocksinna