2010/3/13 Doug Franklin <[email protected]>: > > If you're gonna go apeshit on storage anyway, remember it's cheap. Consider > doing a RAID 10 or RAID 0+1 array to reduce your "OMG I need my backups" > exposure. Keep one or two replacement drives, replace failures immediately, > and you won't have to worry too much. My 4x320GB-drive RAID 10 array has > been running for about 15 months. I had to replace two drives in the first > four months, and none since. No down time, no lost data. As long as only > one drive dies at a time, you don't lose anything, and the firmware can fix > up the new drive in the background, so it doesn't even get in the way.
uhm please take it from a certified hp storage engineer that mirrored stripeset RAIDs - called by the various manufacturers 0-1, 0+1, 1-0, 1+0, depending on their technical flavors, or for the sake of simplicity RAID 10 in the following - has both unnessarily high slack, i.e. difference between phyiscal space and space available for use, and unnecessarily low failure tolerance. I will not go into technical details here unless asked but let me put it this way: RAID 0, 1 and 10 all have n/2 slack, i.e. you lose half your disk space for redundancy no matter what. RAID 0 and 1 are limited to two disks each, hence, RAID 10 is limited to four disks. RAID 0 will not tolerate disk loss. RAID 1 will tolerate loss of one disk. RAID 10 will tolerate loss of one disk under all conditions and loss of two disks only if no stripe is broken so you're playing va banque if you do not immediately replace a failed disk in a RAID 10. That means purchase of a spare disk, online or offline is a should or must. RAID 0,1 and 10 are all non extensible arrays whereas RAID 5 and 6 are both extensible by adding as many disks as your controller can take. RAID 5 has n-1 slack with a minimum disk count of 3, i.e. you lose the space of one disk but never more than one third of the physical space you paid for. The quota improves every time you add another disk. RAID 5 will tolerate the loss of one disk so again a spare disk is a should or must. RAID 6 is like RAID 5 except in that it has a minimum disk count of four and n-2 slack but tolerates the loss of two disks. Data recovery from a faild RAID 5 or 6 is much easier than from a failed RAID 0 or 10 so there is a difference in theoretical risks. Summary: RAID 0 and 10: highest risk, highest space and system cost. RAID 1: moderate risk, highest space and moderate system cost but pointless as opposed to having two USB disk in site rotation and a copy/backup routine to match. RAID 5: lowest space cost, 2nd lowest system cost, 2nd highest security RAID 6: 2nd lowest space cost, 2nd highest system cost, highest security Hope this helps Cheers Ecke -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
