The summary.
RAID IS NOT A BACKUP
It protects against drive failure, not deletion, corruption, ransomware or disaster.
Rebuild is the dangerous period: every remaining drive is read fully under load, and drives from one batch tend to fail together.
VERIFY THE CONTROLLER BATTERY
A failed one disables write caching and performance collapses without warning.
Hardware arrays may be unreadable on a different controller model — keep a spare for anything critical.
MEASURE LATENCY AND OPERATIONS PER SECOND, NOT THROUGHPUT
Typical workloads perform many small operations, and that is what determines how they feel.
SEPARATE STORAGE TRAFFIC FROM GENERAL NETWORK TRAFFIC
Congestion causes timeouts that look exactly like drive failures.
And configure multipathing — without it a single cable or switch failure disconnects storage entirely.
ALERT ON CAPACITY WELL BEFORE EXHAUSTION
Adding storage takes time, many systems slow badly when nearly full, and running out causes an outage.
