raid question

[willawake]

gonna install a new server. got 4x 10.000 rpm 72gb scsi maxtors to be on hp smartarray 6i controller. raid-5 (3+spare) or stripe and mirror using all 4?

cant make up my damn mind. 12 users

[Topcat]

Re: raid question

gonna install a new server. got 4x 10.000 rpm 72gb scsi maxtors to be on hp smartarray 6i controller. raid-5 (3+spare) or stripe and mirror using all 4?

cant make up my damn mind. 12 users

Mirror with spares. For a server, striping is useless. You want redundancy over speed.

[willawake]

Re: raid question

nah, you take two disks and stripe them then you mirror the stripe onto another two disks.

[Topcat]

Re: raid question

nah, you take two disks and stripe them then you mirror the stripe onto another two disks.

0+1 array?? Have you ever tried recovering & rebuilding one of those after a failure?!? That's why I say don't do that, its not fun. Since its SCSI, RAID5 is always an option. I've never had a RAID5 fail on me, so I don't know how difficult they are to recover. Never build a RAID10, so can't help ya there.

[willawake]

Re: raid question

0+1 array?? Have you ever tried recovering & rebuilding one of those after a failure?!?

thanks, good info

that is the problem, with experience in fails you know what to choose. i read so much shit about raid-5 i am getting worried. a friend in uk says forget about what analysis you read, on good current controllers there is no problem. others say with 3 disks and 1 spare in raid-5 performance is not good. i need them to have one volume 72gbx2 cos they have a lot of current data, digital photos etc.

[Topcat]

Re: raid question

For data security, go with a simple mirror. Recovery and RAID rebuild is a snap. If one of the drives in the mirror fails, you can even plug the other into a standard (non-raid) controller, and access as you would a normal drive. A RAID5 would not give that luxury. SCSI 320 is hellishly fast anyway I have it myself, config'd in both multiple arrays 1 mirror and one stripe. The performance difference between the two is noticable on benchmark utilities, but not extremely noticeable in real world performance. Both arrays are f'ing fast!