Quite a packed day at the Summit again. In retrospect I am finding that there might be a method to my crazy/subconscious/random selection of sessions to attend: at work lately my primary responsibilities have really settled into two main areas: HA/DR design and implementation (server and storage planning, and builds) and working with the group that just started our in-earnest implementation of the MS BI stack above and beyond our existing relational warehouse. So at the Summit I have been on a zig-zagging path between HA/DR sessions and the BI track.
Today started with the keynote (more genuinely great new features, and the CTP!) then SQL CAT’s Securing BI Applications in a SharePoint Environment. That session -- my editorial -- could have been called Unbelievably Complicated but Possible: Making Authentication in SharePoint-hosted BI Work, Against All Odds. But Carl Rabeler and Adam Saxton made it clear both that they have tackled just about every permutation of SharePoint and related services, and that they have found a way to make almost all the elements talk to each other. This is one complicated area I know I can’t retain in my head, but there are online resources when the time comes to put this to use.
Next I had lunch with the members of my own team who are attending, which was fun as we had a recap of what people had seen and what they’d picked to attend next. Seems like they are all getting value from the Summit.
I got a better handle on Kerberos by spending yet more quality time with Adam Saxton. It is pretty simple once you understand how things actually work with Kerberos. That was refreshing, especially after the SharePoint war stories from the morning. I’ll have a lot more confidence dealing with this set of issues now.
Next I veered back over to the BI track with Rushabh Mehta’s session Building an ETL Automation Framework. He presented, among other things, a design pattern that can almost automatically roll back erroneous inserts from a bad load to a type II dimension – quite a neat trick. This content should be very helpful if/when we can move our existing BI ETL process over to SSIS.
Lastly, the mind-bending, always entertaining Bob Ward did yet another 800-level session (if there were an 800 level, his sessions would be in it, I’m sure) on internals, this time on latching. I had studied concepts around multi-threaded programming when I did some of my CS academic work, and the need for locking and synchronization to ensure multiple threads don’t corrupt data. But I wasn’t clear exactly, in SQL Server, the difference between locking, latching, and which one or both provided that protection. Bob did a great job of explaining that, and it was very illuminating. I won’t be running the debugger, or breaking out the calculator to convert bit maps from hex to decimal (I hope), but I love knowing more about how this stuff really works. Good time.
I expect everyone else is at Gameworks as I write this, but I am lucky enough back home, got to read stories to my kids before bed.