Sheesh, why is this guy's blog post getting so much attention all of a sudden? It's been out for months; suddenly, everyone's talking about it. I'll copy what I wrote in last week's thread on it:
On Fri, Oct 16, 2009 at 7:23 AM, Ben Scott <[email protected]> wrote: > As I understand it, the argument is: Given a 12 terabyte disk array, > reading all of the blocks from all of the disks in the array will mean > you've read enough data to reach the URE rate (Unrecoverable Read > Error) quantity from the spec. Since a RAID 5 rebuild means reading > all of the blocks in the remaining good disks, you'll hit a URE and > the controller pukes. This is an issue now because while disks sizes > have increased, the URE rate spec has remained constant. > </restatement_of_argument> > > One thought I have: Most RAID controllers, for the past few years, > include a "Patrol Read" function that regularly reads all of the > blocks from all of the disks in the array. The author's argument > would suggest that if you have a 12 TB array, it would immediately > start failing the Patrol Reads from day one. That doesn't seem to be > happening. Why not? (While I don't have a 12 TB array, I know others > do.) -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/> ~
