We have yet another post this morning. This one is from Adam Machanic, who will be presenting A Day of Doing Many Things at Once: Multitasking, Parallelism, and Process Distribution - you can find more about his session here: http://sqlpass.eventpoint.com/topic/details/AD311P.
Is there an audience that would benefit especially from this session?
As database professionals, we must learn to adapt to change. And it is easy to see that our world is changing, rapidly. As recently as five years ago, I/O performance was the name of the game -- almost to the exclusion of everything else. Since then, CPU power has rapidly multiplied, memory capacity has exploded, and solid state devices have become commonplace in the data center. All of which means that I/O is simply not as important as it once was. Data sizes continue to grow, so even as the I/O system delivers data more quickly -- or that data is read directly from the massive in-memory cache -- many queries don't seem to get any faster. The reason? Systems generally fail to consider the new bottleneck: processor resources. This seminar will teach you how to use CPU power to your advantage, making those slow queries scale to meet the demands of your end users.
The seminar is geared toward developers working with large sets of data -- queries that deal with millions or even billions of rows -- but there will also be plenty of material of interest to production DBAs tasked with managing these large SQL Server instances. And even if you don't work with large data today, you eventually will; there is no avoiding the push toward bigger and more interesting databases. (And really, why would you want to avoid it?)
After having attended your seminar, what are two or three things that an attendee will be able to take back to the office and put to use right away?
After attending the seminar you will have a solid understanding of how to read and interact with parallel query plans. You will understand why and how performance is impacted by parallelism and how to control it using both plan shape and server options. If you need to tune large queries on modern hardware (servers with 8 or more cores), you will return to the office with a number of tricks and tools that will help make your query performance both better and more deterministic.
What background should attendees ideally have to be fully prepared for your seminar?
It will certainly help if attendees have spent some time studying the basics how to read query plans. The parallel aspects will be covered in depth, but I won't take too much time to explain the underlying non-parallel aspects. A background in computer science fundamentals will help with some of the theory I'll be discussing, but I'll assume that most people in the audience haven't studied CS and will fill in the gaps as needed.
What experience are you, as a speaker, bringing to this session?
I have worked with SQL Server for 11 years, and for the past 8 my career has been focused entirely on SQL Server development. Most of that time has been spent working with large databases, and performance optimization is always at the top of my list. As a consultant I have seen dozens of environments and have been tasked with tuning a huge variety of queries, including many that work with billions of rows. Over time I have developed patterns and tools that have helped me achieve the best possible performance from these monster data consumers, and this set of techniques is what I will share in the seminar.
You can register for the 2010 PASS Summit here: http://www.sqlpass.org/summit/na2010/Registration.aspx
See you in November!