The Big Red Button: How to Use sp_Kill

Production is on fire, users are screaming, and someone is hovering over the server restart button. This quick talk is about what to do before that happens.

Brent Ozar introduces sp_Kill, a new open source stored procedure designed for performance emergencies. It's designed to help when runaway queries, blocking transactions, or forgotten sessions are dragging the entire server down. This isn’t a tool for careful, glove-on analysis. It’s built for real-world scenarios where you need to act quickly, safely, and with guardrails.

Brent will walk through common emergency situations: bad query plans, sleeping sessions with open transactions, jobs running at the wrong time. He'll show how sp_Kill helps identify what’s going wrong, log everything for later analysis, and guide you toward the least risky sessions to terminate. You’ll see how sp_Kill balances speed with safety, why blind killing is dangerous, and how this approach is still far better than restarting SQL Server out of desperation.

If you’ve ever wished for a big red button that’s smarter than "Restart", this session is for you.

Who sp_Kill is for (and who it's not for) How to use the most common parameters How to set it up so your help desk can run it