Our pre-production environments -- Test, Stage, and Preview -- are housed on a single server. Stage has been significantly slower than the others for as long as I have been here (two years). Even during off hours, with virtually no load. Particularly, it takes much more CPU than the others. Memory and disk I/O seem ok.
Things I have looked into so far:
- The schemas are nearly identical. Preview is synchronized with Stage before each rollout, which is about every 4-6 weeks.
- The data is also nearly identical. The DBA uses a backup of the newly-updated Production to refresh Stage and Preview aftear each rollout.
- I have reorganized/rebuilt a handful of heavily fragmented indexes.
- Instance settings are identical. There is no prioritization, either.
- Overall health of the server seems pretty good. Test and Preview work about as efficiently as you'd expect, considering they share an underpowered server!
So, any ideas why one instance's queries crank the CPU while the others seem ok?