Presentation 5: Presenting Performance Tuning with Memory Grants

Arthur-Daniels, 2019-04-15 (first published: 2019-04-01)

So it’s about a week after SQL Saturday Chicago 2019. The event was amazing, tons of great people and in a great venue. Here’s the details of what happened between Presentation 4 and SQL Saturday Chicago.

Final steps to prepare

Between the last post, presentation 4, and that day, I refined a lot of the presentation. Then I presented it to my team. They were very supportive, which was comforting. I also learned what parts I would need to practice.

While practicing on Friday evening, I found that my system would freeze when I ran my resource semaphore demo. Uh oh. There wasn’t enough time for me to completely debug the demo, although I could see that when it happened, my computer’s hard drive was at 100%.

Rushed for time

Resource semaphore waits are really important for my memory grant presentation. They’re the big problem that’s caused by excessive memory grants, and I wanted to at least show what it looked like. However, with only a few hours remaining, I had to find an alternate solution.

I decided to capture a screenshot during the issues, as a way to prove that the issue was occurring. Even if my system wouldn’t respond, I could show the screenshot.

Here’s that screenshot:

SQL Saturday Chicago and my presentation

Then, on Saturday 3/23, I presented in front of a full room. I was really happy to have so many attendees, and the room had a lot of questions.

And, as expected I needed to use the screenshot. Not good! But it’s better than nothing, and I could show in my later demos that the queries would run with the improvements I suggested.

Overall thoughts

The presentation itself was a success, despite the demo issue. I have time now to do some troubleshooting on my laptop, and I have the screenshots for the next time I present.



Original post (opens in new tab)
View comments in original post (opens in new tab)