Another year, another October trip to Seattle. As the rain falls across the city, database professionals descend upon the Downtown area for the annual Pass Summit. The Pass Summit is one of the biggest database conferences in the world, offering industry-leading speakers, in-depth training, technical insights, and is attended by data novices and pros alike.
This year, I’m presenting two sessions at the Pass Summit: TempDB and Query Store.
The first is a deep dive into TempDB, looking at what uses TempDB and why. Then looking at two kinds of contention that can be seen with TempDB; IO contention and allocation contention. IO contention comes from driving the IO subsystem harder than it is capable of, resulting in slow response when reading or writing to disk. Whereas, Allocation contention comes from SQL’s file structures, and occurs when tables are being created at a high frequency, resulting in processes waiting to access key allocation pages in the database. The two different forms of contention require very different approaches to resolving them, and mis-identifying the problem can result in a lot of wasted time.
The second presentation I’m giving is on the Query Store, a new feature in SQL server 2016 that promises to change the way query performance monitoring is done. This presentation looks at how Query Store is configured, how it works, and shows how to debug a query performance regression using the Query Store’s built in reports.
In addition, a group of MVPs are putting on a 1-day ‘freecon’ on the Monday before Pass Summit, for those who aren’t attending a preconference session but still want to spend the day learning. I’m presenting there as well, specifically, a session on misunderstood wait stats and why some waits are not as concerning as they appear to be.