pattern // automationPRODUCTION
Operational Baselines Before Alerts
Know normal runtime, row count and cadence before deciding what “unhealthy” means.
Trust & provenancePRODUCTION
Evidence tierPRODUCTIONOfficial sources0Other sources0Last verifiedNot recorded
Evidence tier describes my hands-on relationship with the content. Official sources describe platform contracts/documentation. Production observations should not automatically be treated as official limits.
Problem
A green status can hide slow degradation, while arbitrary thresholds create noisy alerts.
Approach
Capture expected row-count ranges, typical Automation/Query runtime and freshness windows. Alert on meaningful deviation from that baseline, not only hard failure.
Requires a short period of observation
Thresholds need periodic review as volume changes