I measure High Frequency cached statements with decimal executions-per-second arithmetic. A customer wanted to identify the busiest queries.
SQL SERVER – Using NOEXPAND with Indexed View
NOEXPAND treats an indexed view as stored indexed data instead of expanding its base tables. Customer results have differed.
SQL SERVER – Parameter Sniffing and Bad Plan
Disabling parameter sniffing does not guarantee that a bad plan becomes good. That remains my answer to the customer question.
SQL SERVER – Dirty Pages Clean Pages and the Correct Cache Commands
Clean Pages have no outstanding data-file write, while dirty pages contain unwritten changes. A customer asked which state was better.
Status Changes Only: Collapsing Repeated Rows With LAG
Hundreds of repeated readings, three real changes. Here is how LAG finds them, and why a plain not-equal test quietly drops rows.
Query Progress: Watching a Running Plan With dm_exec_query_profiles
Elapsed time cannot tell you if a query is almost done. Live operator rows can. See how to read them and how far to trust the percentage.
Tuning Query Cost 100% – SQL in Sixty Seconds #181
After the blog posts, lots of people asked me to build a video so here is the video discussing the topic of Tuning Query Cost 100%.