[ Date Index ][
Thread Index ]
[ <= Previous by date / thread ] [ Next by date / thread => ]
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Robin Cornelius wrote: > > PS i wonder what the actual probability is for a double disk disaster? I > suspose you would have to factor disk life, mean time between failures and > the actual normal distribution of failure for a given disk which i suspect > the manufactures would not freely give away. Nope, wrong approach. You are assuming disk failure is spontaneous due to aging of the disk drive (you'd also have to factor in mean time to repair, since I've seen two RAID systems die due to a second disk going before anyone had fixed the first - some RAID arrays use a hot standby to help avoid this). The key point is that disk failures are almost invariably due to other factors that affect multiple disks at the same time, the classic case being disks that fail after a controlled power down, either heat shock, or power surges during boot (which your UPS won't save you from). Double disk failures in well designed and well maintained systems are still rare, but I've seen reports of coffee in computer rooms against "rules", and also people tripping over and falling on disk arrays, achieve it. The Met Office famously had an engineer (I use the term loosely) rewire the mainframes three phase AC wrong and take out a whole series of disk arrays. So the model must be based on experience to factor in real world events, not just on the abstract numbers disk drive companies quote. Remember also most disk drives quoted MTBF is based on a large number over a short period, and probably isn't accurate for a small number of disks running for a number of years when wear and tear become an issue. Interestingly when we did some work planning for serious disasters, any computer system lasting 3 to 5 years, even under near ideal conditions, the odds on it being "wiped out" are still of the order of a percent or two. Most of the time life plods along nicely, but then one of the four horsemen of the apocalype gets bored waiting for Armageddon. -----BEGIN PGP SIGNATURE----- Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQE+9HX2GFXfHI9FVgYRAs48AJ9cuZlq+e0+xrFMCPfCLQVdarqHmACdG+RE 0Snd5WAUigfEKHDur7oplVQ= =0/LL -----END PGP SIGNATURE----- -- The Mailing List for the Devon & Cornwall LUG Mail majordomo@xxxxxxxxxxxx with "unsubscribe list" in the message body to unsubscribe.