On 27/01/2021 13.04, Per Jessen wrote:
Carlos E. R. wrote:
On 27/01/2021 12.49, Per Jessen wrote:
Carlos E. R. wrote:
On 27/01/2021 10.45, Stakanov wrote:
A question about "battery": is it still an argument if you system is backed up by an UPS? I thought the batteries on controller cards were paramount only on systems that are not protected by a convenient UPS. Wrong?
The idea is, I assume, that battery backed hardware raid detects that the power died and commits everything to disk before finally powering down the HDs.
The battery is there to power the write cache memory until mains power returns. These days it is flash backed cache, not battery backed.
I guess that cache has a significant size?
It depends - in earlier days 64Mb, 128Mb, 256Mb, today 1024Mb. (maybe more).
Let me see. A "Seagate BarraCuda 3.5" 4TB SATA3" has an internal buffer of 256MB. Then those cards don't have a "significant size" of memory, IMHO. And begs the question about who backs up the hard disk buffer memory. Suppose the computer sends 3 write operations and then loses power. The first operation completes to the "rust". The second is waiting in the internal buffer of the hard disk, the third is still on the card buffer. Just a "suppose" situation. On power restore, operation 2 is lost, but operation 3, out of sequence, is applied. Can be a disaster... so what, disable the internal disk buffer? Can the card buffer be as efficient as the internal disk buffer? I doubt it. Best thing would be for the disk to also have a backup battery or capacitor. -- Cheers / Saludos, Carlos E. R. (from 15.2 x86_64 at Telcontar)