Am Montag 08 Dezember 2003 02:45 schrieb Xose Vazquez Perez: > Allan Metts wrote: > > This is an Adaptec AAA-133U2, apparently part of Adaptec's "Array 1000" > > family. > > > > --*-- Anyone see any issues with this controller and Fedora support? > > is it an aacraid board ? > > > --*-- It sounds like the general consensus is that, with five disks, I > > should use RAID 0/1 with a hot spare -- provided I can live without 60% > > of my disk space. If not, use RAID 5. Everyone agree? > > raid 10 or 01 = 1/2 of total storage > raid 5 = total_storage - 1_disk > > but raid 5 only supports a failure in *one* disk, and the rebuild of the > raid is _very_ slow. There is big degradation if one disk is broken. > Raid 10 can supports 2 failures in differents set of raid_1 and degradation > with 1 disk bad is less than raid5. > tht is not correct. when one disk in each arry fail the all data are lost. only 2 disk or more in the same arry can fail. You are describing the failure behavior for RAID01 which is two stripes (RAID0) which have been mirrored to each other. The recommendation was/is to use RAID10 which is N (in this case two) mirrored pairs (RAID1) which have been striped (RAID0). In the case of RAID01, yes, if any two drives on the same strip fail you are in trouble (though data is not neccessarily lost, one could take the mirror and stripes apart and rebuild one stripe from the remaining drives, but that's not the point). In a RAID10 array, on the other hand, since the stripe is assembled from mirrored pairs - unless both drives of the same pair are lost the array can continue to function, provide data, and be fully recoverable with as many as half of the drives having crashed. Since the most common cause of multiple drive failure is a bad manufacturer's lot of drives (I've experience this myself) you can minimize the possibility of data loss by being diligent to construct each mirrored pair with two drives from different manufacturer's lots. Art S. Kagel
