Query Optimization Tops David DeWitt Keynote Voting

You voted, and the results are in: Dr. David DeWitt will be speaking on Query Optimization 101 during the most anticipated keynote of PASS Summit 2010. The query optimization topic attracted 61% of the votes over MapReduce vs. SQL (although one person on Twitter asked, "Can't we have both?").

“In two short years, giving a keynote at the PASS Summit has become a highlight of my year,” noted Dr. DeWitt, Technical Fellow with Microsoft’s Jim Gray Systems Lab. “This year, I am excited to have the opportunity to talk about SQL query optimization, why it is hard to always produce good plans, and new technologies that offer the promise of better plans in future releases of SQL Server.”

Thanks to everyone who voted! We hope you can join us live in Seattle in just a few months to enjoy the keynote in person. In the meantime, you can read more about both topics below and click here to rewatch Dr. DeWitt's wildly popular PASS Summit 2009 keynote.

Winning Topic: Query Optimization 101
Query optimization is the process of compiling a SQL query into an executable plan that is as efficient as possible. While the basic fundamentals for cost-based query optimization were first described in a seminal paper by Pat Selinger of IBM Research in 1979, optimizing complex queries continues to be very challenging 30 years later. In this talk, David DeWitt will describe the fundamentals of query optimization, why it remains a very hard problem today, and present several recent research results that promise to improve the quality of plans produced in the years to come.

Runner-Up: MapReduce vs. SQL
Over the last several years, many companies have begun using MapReduce for analyzing large data sets. First developed by Google, the open source version of MapReduce known as Hadoop is capable of running on clusters consisting of 100s to 1000s of low-cost commodity processors. This talk will describe what MapReduce is, how it works, and how it can be used to analyze large data sets. David DeWitt will also compare and contrast MapReduce with the capabilities provided by a database system such as SQL Server and will describe how the two can be used together effectively. Finally, he will present the result of a set of benchmarks comparing the performance of MapReduce with a popular parallel SQL DBMS running on a 100 node cluster.

Gold Sponsors