LearnDaily Operations
How To Find Out What Changed Overnight
1 min readUpdated 6 Nov 2026
A query that was fast yesterday and slow today is rarely spontaneous. Something moved: a schema, a permission, a setting, a job, or the data itself. Here is how to find it fast.
Where changes are recorded
- The default trace: schema changes, permission changes, autogrowth events and some configuration changes.
- The error log: instance restarts, startup issues and serious resource messages.
- Job history: schedule edits, new jobs and jobs that ran longer than usual.
- Deployment and release records: application changes that changed query shape.
- Audit or DDL triggers, if the environment already has them.
Changes that do not look like changes
Some of the most disruptive changes are recorded nowhere, because they are normal operations for someone else.
- Statistics updated, so a plan flipped.
- A database grew, so the file allocation changed.
- A new database was added and no backup job covers it.
- A login was granted a broader role for 'just this week'.
Build the answer before you need it
- 1.Keep a running list of what changed in the last 24 hours, per instance.
- 2.Keep the error log and job history within reach instead of digging for them.
- 3.Compare today's top waits with yesterday's, so the change shows up as a difference.
A change log is only useful if it is already there when the incident starts. Reconstructing it afterwards costs the hours you were trying to save.
What to remember
- Check the default trace, error log and job history first.
- Statistics, growth and new databases are changes too.
- Compare today's waits against yesterday's.
- Keep the change list running, not reconstructed.
