High-Risk?: RAID5 sets over 1TB...

What does everyone think about this...

At my workplace, we have a server that has a bunch of disks in it, assembled into a RAID5 set totaling over 1TB. One day, we came into work, and the server had crashed, with a double disk-failure. From the logging we could find, one drive failed, then some time after that (over a weekend), and 2nd drive failed, killing the system.

"Of course it will fail ", one guy said -- he said that when your RAID5 set is that large, and a drive fails, it's just too strenuous on the remaining drives, "I've read articles on it..." over the next week, we try rebuilding about 3 different times, and had the same failure - double-drive failure on a RAID set over 1TB.

Question: Did we just happen to have a bunch of spare drives that were bad, or is there any validity to these statements.. Large RAID5 sets are high-risk, and don't re-build well?

Does it have to do with the disk-type? Or the HW controller? I've got a 10TB RAID5 set running on a Sun SAN (usual 1TB SAS disks), and it runs fine. If I wanted to, could I run out and buy 6 two-TB disks, and a controller, and set up a RAID5 set comprising 10TB if I wanted to, or is that certain death?

 

Hard disks would fail no matter what. It doesn't matter whether they are in RAID 5 or RAID 0. Size of the array has got no direct impact on the reliability.

SAS disks are built better & are better suited to critical applications. But even they would fail at some point thats where backup comes in. And having one or two extra disks in array as failovers is pure bliss & might save your behind.

 

this confirms what I thought. --I didn't think that the size of the RAID container had an influence on whether or not it will actually rebuilt. I "m not sure about the HW specifics, but the comment that was given to me internally here was more conceptual, than related to HW, so I left out those details.

Thanks!