Finding the Limits

Having conquered the bulk API and hit 16M rows/sec, we turn our eyes to other workloads to see just how hard we can push SQL Server before it breaks. In this session, I will take simple SQL statements, the stuff you write every day, and bump up the scale until things start breaking. I will attempt to answer questions like

  • How fast can you read rows from SQL?
  • How many INSERT statements can you run per sec Is UPDATE really so bad? How does index build scale? And how fast can you build a Denali column store? Just how many IOPS can you squeeze out of a modern machine running Windows? (Give those SAN guys something to think about)  How fast can you table scan from disk? (and how do you test it?)

The session will also introduce tools like WinDbg and Xperf for troubleshooting.

Presented by Thomas Kejser at SQLBits IX
  • Downloads
  • SpeakerBIO
    Thomas has been working with SQL Server since version 6.5 and performance tuned cubes since OLAP Server 7.0. He has worked on some of the largest cubes in the world. Performing fast loading of warehousing and removing every bottleneck is one of Thomas specialties. He holds a master degree in computer science and has 10+ years of experience with developer coaching and scalable architectures on both OLTP and OLAP systems.
  • Video