Network Computing is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

On Large Drives And RAID: Page 2 of 2

Systems like 3Par and Xiotech's Emprise that virtualize RAID data and spare space across all their drives should rebuild a big drive faster. Using 2TB of spare space across many drives rather than a spare drive eliminates the bottleneck at the single spare.

Even if another drive doesn't fail, the rebuild may fail when it can't read a sector from one of the remaining drives in the array. A 2TB drive holds 1.6x10^13 bits and has a published error rate of 1 in 10^15 bits. Rebuilding 1 4+1 RAID-5 array means reading and/or writing 8x10^13 bits resulting in an 8% chance an error will occur during the rebuild. That would result in either a rebuild failure or a message in the array logs that block 32322 of LUN 2232. This is bad depending on how gracefully the controller handles the error.

Switching to mirroring reduces the odds of data loss to 1.6% per rebuild but that's not good enough.

The moral of the story, dear readers, is big drives require some sort of multiple parity or RAID-6 system, especially if you're using them to hold deduped data that could affect many data objects.