[snip a bunch of talk about why disks fail, blah blah..] a quick hard drive failure primer:
ATA drives have a MTTF (mean time to failure) rate of between 500,000-1,000,000 hours, and SCSI drives have lots more. some manufacturers publish these, some don't - but when they are published, typically they don't include what's referred to as the "duty-cycle", which is basically the percentage of time that the disk is 'active' in the context of their MTTF ratings. ie. if you have a disk with an MTTF of 500,000 tested at a 100% duty cycle, then it's equivalent to a 1,000,000 MTTF tested at a 50% duty cycle. (and no, they don't actually run them for 500,000 hours..) the three things that kill disks: usage, heat, dust (no, disks aren't vacuum sealed), vibration aside from usage, the most important one is operating temperature (factoring in the arrhenius curve - w00t, i found an excuse to use the name of a dead swedish guy) and ambient temperature. and then you put these things into a bath tub curve, you get negative aging (infant mortality) and positive aging (wear-out) but in the base of the tub you usually end up with a failure rate of 1-2% per year. power supplies killing disks? um... never heard of any data supporting this - it's usage, heat, dust and vibration. -josh > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On > Behalf Of Gregory Hill > Sent: Thursday, May 11, 2006 1:47 PM > To: Provo Linux Users Group Mailing List > Subject: RE: Home RAID question > > I'm going to have to assume you work in an IT department or something. > I haven't RMA'd a hard drive since the big IBM fiasco about 5 > years ago where they released a ton of faulty drives. I've > had dual WD 74GB Raptors in RAID0 for about 2 years with no > problems (using an onboard promise sata raid controller, then > the nvidia sata raid controller after that). I've probably > only RMA'd 3 or 4 hard drives in my life. Come to think of > it, I haven't had to RMA a hard drive since I started buying > good power supplies. Do you use the cheapo ones? If it came > with the case, then the answer is almost always yes (though > there are a very few rare occurrences of cases coming stock > with quality power supplies). > Would be interesting if there's a correlation there. > > Greg > > > -----Original Message----- > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf > Of > > Corey Edwards > > Sent: Thursday, May 11, 2006 1:30 PM > > To: [email protected] > > Subject: Re: Home RAID question > > > > On Thu, 2006-05-11 at 13:22 -0600, Andrew McNabb wrote: > > > RAID 5 makes it so that a single disk can die without having to > restore > > > from backups. However, it's always possible that two will die at > about > > > the same time. This is especially true if your disks are from the > same > > > lot, or if it takes several weeks to replace a disk. > > > > Or if the disks are in the same computer and you get a > power spike. I > > just want to emphasize how common it actually is to lose 2 disks at > the > > same time. Not every day, certainly, but it's happened to me > personally > > twice in the last two years. In that same period of time I've RMAd > > probably 10 other drives for single failures. > > > > Backups, backups, backups! > > > > Corey > > > /* > PLUG: http://plug.org, #utah on irc.freenode.net > Unsubscribe: http://plug.org/mailman/options/plug > Don't fear the penguin. > */ > /* PLUG: http://plug.org, #utah on irc.freenode.net Unsubscribe: http://plug.org/mailman/options/plug Don't fear the penguin. */
