RAID write times are improved by writing to a cache.   This is a quick operation, the calling program can get on with its business, and the RAID controller can solidify the cached changes to disk(s) in its own time.  The cache power supply has a backup battery.

When the cache backup battery condition measures below a safe limit it is marked as FAILED! (which raises alerts) and the RAID controller stops using the cache and writes only direct to disk.

Performance Effect?

How much does that affect performance?   Those in the know (some network engineers) suggested it might add 20ms to write times.  They weren’t far wrong.  The Disk avg write time metric on the Redgate monitor showed just under 30ms increase

It had no effect on average disk read time, as would be exected: