Keynote:  Wow… I feel smarter now.  David Dewitt, a Microsoft Technical Fellow who works in the Data and Storage Platform Division, presented this morning all I can say is WOW.  If I had professors in college that could speak half as well as David, I would have a couple of PhD’s at this point.  His presentation focused on how to change the way database engines store data based on research he and his students have performed in analyzing bottlenecks of today’s hardware.  He highlighted the bottleneck in today’s architecture is really focused on physical disk IO and how changing the paradigm of how we store data can significantly improve the read performance in data warehouse scenarios.  He further predicted that the database market may segment between OLTP based systems and warehouse / DSS systems.  By the end of the presentation, I was wondering if I should head back to school…

SQLCat  / SQL Server Consolidation and Virtualization Best Practices:  This is a sweet spot area for me as I’ve done plenty of work in this arena.  In fact, I chose to go to this presentation because I had passed one of my white papers to the team that was presenting.  It was interesting to see how the work I had done in analyzing the performance implications of virtualizing SQL Server in Hyper-V had been taken and expanded upon.  My key take away, work very closely with your hardware vendors to ensure you are purchasing hardware that can be fully exploited by Hyper-V.

Advanced Nested Sets model in SQL Server:  I have always enjoyed the mental exercise of reading Joe Celko’s books so the opportunity to hear him speak was an opportunity I couldn’t pass up.  I had seen his work on tree navigation and always dismissed it as a clunky way to manage trees, but after hearing him explain the methodology it made much more sense than I had ever given it credit for.  I think I may need to go back and review some of his other books and see what other design patterns I need to revisit.  I really enjoyed hearing him speak and again, you can easily pick out the people who have experience teaching and speaking at the collegiate level from the rest of us engineers.

Using Policy Based Management to Manage your Environment:  I know… policies are made to be broken.  I wanted to review this class to make sure that I’ve got a strong understanding of policy based management.  I really like the notion of having a set of policies in place that will prevent “rogue” dba’s (or developers who read to many articles) from changing settings that I’ve specified for good reason.  Policy based management allows me to hand over control of a machine and still rest easy knowing that the configuration will not drift from where I expect it should be.  This is really important as you start to manage multiple environments such as a development / test / production solution.  A consistent set of policies will ensure that all users of the system have consistency as they move between environments.  While defining a complete policy solution can be a daunting task, the recommendation was made in the class to build one or two policies a week and slowly chip away at the configuration drift.  Once you have a baseline that you are satisfied with, policies can be treated as normal configuration items that need to be deployed as a part of any change to the system.

Sizing SQL Servers for unknown Workload:  Ah… the final session of the main conference.  This topic helps bring real engineering design to what has become a “seat of the pants” sizing exercise.  The presentation focused on taking a small “test” workload, and profiling its results.  The key is to ensure you have a representative sample of work from the application being profiled.  From there, he presented a series of scripts that analyze the collected profile trace and develops a breakdown of the workload by inserts, deletes, selects and selects with joins.  This information can then be funneled into SQLStress and then help predict the expected performance on the targeted hardware.  An interesting approach to a very complex problem.