On Thu, 2007-08-09 at 09:13 -0400, Neal Becker wrote: > Bruno Wolff III wrote: > > > On Sun, Aug 05, 2007 at 21:42:01 -0700, > > Dave Stevens <geek@xxxxxxxxxxxx> wrote: > >> Hello all, > >> > >> I'm thinking about a software RAID (0) setup for my next system revision > >> and am curious where this occurs in (I assume) the system setup. Can > >> anyone point me to documentation on this? > > > > You can do a custom disk layout during install. It isn't too hard to > > figure out do a raid 0 install, though choosing good values for chunking > > might take some guesswork. > > > > What are you going to be doing that makes you think raid 0 is desirable? > > For modern home machines with lots of memory, this probably isn't going to > > give you much of a speed up in typical use. It will make the chances of > > your file system being lost because of a hardware problem, about double. > > Raid 1 or 10 seem to be more useful for home users, than raid 0. > > > > No! If the prob of failure during an interval T is pf, the prob of failure > for an array of 2 drives during T is 1 - (1-pf)^2. > Just for completeness it should be pointed out that this is for random failures. If the failure mode is non-random, say a design flaw or mechanical assembly error, the formula doesn't hold, and the error probability is nearly the same as a single disk. Although in "real life", the errors are likely somewhere between random and design related, so the formula is very close to the actual figure, especially given reasonable testing standards by the manufacturer. Regards, Les H
