Finding Poorly Performing SQL Queries
dbWatch sponsored on-demand webinar
Before you can tune queries or change application code you have to understand what things are causing pain. Not all slow or badly written queries matter. Reducing a query from 30 seconds to 1 second feels great, but it might not affect perceived application performance much if it only runs once a day!
So how do we identify these poorly performing queries? A good monitoring solution can help because you can see the history, but we can find plenty of pain using Profiler (old school) or xEvents and that will be the focus of this presentation. We'll look at how to capture the data and what filters will be useful to find long-running or high-resource queries. We'll also look at how to know when blocking is the culprit versus the query itself and how to dig into deadlocks. Then we'll talk about prioritizing; identifying easy wins, which ones you can fix, and which ones need a developer and/or a change in approach.
Once you master this approach of finding poorly performing queries you have something you can measure and repeat. Find 10 poor queries, fix them, measure the improvement and then repeat. It will absolutely make you feel more in control of managing and explaining the performance stance of your servers.
Speaker - Andy Warren and Per Christopher Undheim
To access materials please fill out the form below.