Tag Archives: SQL Pass Summit

Fourth day at #SQLPass

Started early today with a breakfast seesion at 07.00 with “SANless clustering with FusionIO and SISO”. Interesting session where I learned about a limitiation in AlwaysOn that id didn’t know of. Apparently only one thread per Availability node is used for SQL replication between nodes in the cluster, limiting throughput to about 40 Mb/s. That could potentially become a problem in a larger environment. Think about Sharepoint….

SISO clustering seemed like a nice build-on feature to Windows clustering, combined with FusionIO it looks like a great solution which seems very simple to install and maintain. Need to check pricing on that.

Next was a wonderful session with Kimberly Tripp on Statistics and skewed data. She have put together a fantastic collection of Stored procedures to find and create statistics for skewed tables. I really love to come home test this on my customers 11.000.000.000 row table which have some strange query performance from time to time. I bet it will solve a a lot of problem. I think they will be available on hhtp://www.sqlskills.com website shortly.

After lunch Rob Farley from Down Under hosted a session called: “Theses aren’t the joins your looking for”, where he deep dived into the different forms of joins and what kind of actions they have on returned rows. I won’t even try to get into details, but he a great presenter with a good deal of jokes and conversions with the audience. It was very entertaining.

The CAT team had the put together a session where they compared performance on ROLAP vs. Tabular when running on Parallel Data Warehouse. This session got a bit outside of what I had thought it would be about. Actually the first session so far where I totally lost concentration and started thinking about other stuff. The topic was interesting even for me who is not a BI developer but the presentation skills where not that good and to many really weird questions kept popping up all the time. Too bad, I wanted to learn more about PDW.

Last, Paul Randal is on stage to talk about waits, latches and spinnlocks. This was a great level 400-500 session on deep internal stuff. He is really awesome and his presentation got 25 minutes late. No one cared, it was so interesting to see how to pin down contention on this level. I think I have a couple of environments back home where I can actually have use of thhis new knowledge, can’t wait to get home…

Evening finished up in Nasacar Hall of Fame with the Community appreciation party. Kind of cool place where you go a interactive tour to prepare yourself and your car that you build during the tour, to finally drive a NASCAR simulation. Given that we where like 3-4000 people I skipped that part and went for food and beverage.