A Self Tuning Fill Factor Technique for SQL Server

Presented by: Mike Byrd

Have you ever created an index and wondered what to specify for fill factor? Usually I’ve never ever given it a worry and just left it at the default value = 0 (actually this is a synonym for 100). But recently I’ve stumbled on a technique that will determine a “near” optimum fill factor value for every index in a database. I say “near” because it won’t always derive the perfect fill factor, but will get you close.

We will look closely at the TSQL for this technique to see how the proof of concept was developed for a very active OLTP production database. I’ve observed a decrease in overall database wait times of 30% after the tuning. The tuning is continuous and allows for changes in data skew as well as evolving application interactions with the database.

This session was part of GroupBy Americas in the Database Administration track.
Stay updated on GroupBy’s next online conference on October 27-28, 2020: https://groupby.org