1

I mean, beside any incident into the rack, is possible that a redundant storage like the hp msa1040 can die because of a single point of failure?

I'm asking because we are going to implement a redundant virt system with 2 servers and a san (let call the msa1040 a san), but I am quite new to the topic and I don't see a reason why such a san (sas attached) could die.

Is in any single point of failure in such kind of systems or everything is really redundant in them as advertised?

Thank you

Edit: for downvoter, it can still be a dumb question but I'm aware that everything can die. I was wondering if there actually is a non redundant component in such items. Then yes even redundant components can blow altogether but to me these are not single points these are overall system failures caused by failures of the given redundancy subsystem. It is a different thing.

EDIT 2: while I've marked the answer as the accepted one, I would like to point out what, to me, is the best answer in absolute: https://mangolassi.it/topic/8822/why-dual-controllers-is-not-a-risk-mitigation-strategy-alone

matteo nunziati
  • 624
  • 1
  • 4
  • 13

1 Answers1

1

Yes, any storage array can stop working for a reason or another. The RAID configuration can fail (see that question about RAID level), the backplane can die, etc...

Be sure to have good backup and a recovery plan.

By recovery plan I mean: If you use in example a backup library and it's a VM that use it. Even if you have good backup, please consider the time to bring back the array + time to restore a VM and reinstall the backup software + time to re-inventory the cartridge + time to actually restore. If unplanned for that step, in real life, as I seen a SAN crash (the array got toasted, it was in RAID50 and sadly two disk in the same set died in a short time lapse), it took 3 days 24/7 to recover.

So for your actual question, yes it can fail (so don't put your eggs all in the same basket)

Community
  • 1
yagmoth555
  • 16,300
  • 4
  • 26
  • 48
  • Yes backup was of course considered. I was wondering is therecwas qcsingle point of failure and you suggest raid failure being the most relevant. So no single mobo on the storage wich can fail or similar? Mostly a raid failure as a cause? I was thinking about non redundant components. – matteo nunziati Feb 03 '17 at 06:32
  • 1
    @matteonunziati if its a dual controller unit, only the backplane is not redundant I think, but that rarelly fail. greater risk with the disk in my own opinion – yagmoth555 Feb 03 '17 at 13:11
  • 1
    @yagmoth555 I'd agree with that - backplane failures are rare, and even then, on the same physical backplane the redundant data paths are completely isolated electrically. I'd add that in my experience the most likely problem on such storage units outside of disk failure is a person actually doing something *wrong* that winds up breaking things - such as plugging the redundant power supplies into the same power source, plugging in the connectors wrong, unplugging the wrong connector, or misconfiguring the device. In other words, worry more about the admins and techs than the device itself. – Andrew Henle Feb 03 '17 at 13:34