Re:Official Google Blog: Sorting 1PB with MapReduce
by russell on Saturday November 22, 2008 @01:52PM
(#373070)
Not at all surprising given the large number of disks involved. Here is a very interesting analysis of actual Google failure rates: http://storagemojo.com/2007/02/19/googles-disk-failure-exper...
Re:Official Google Blog: Sorting 1PB with MapReduce
by nebula on Saturday November 22, 2008 @03:22PM
(#373160)
Thanks for the pointer; it's interesting.
Re:Official Google Blog: Sorting 1PB with MapReduce
by wmf on Saturday November 22, 2008 @02:32PM
(#373102)
1 failure per 50K in 6 hours is the same as 1 per 34 in a year, or a 3% annual failure rate. I think this is the same number Google reported in their disk failure paper.
Re:Official Google Blog: Sorting 1PB with MapReduce
by nebula on Saturday November 22, 2008 @03:41PM
(#373173)
Actually, I did this math before I posted my comment.
What I don't understand is this:
With x% annual failure rate, are the failures uniformly distributed over time? I would expect that it's probably a normal distribution over time. (i.e., no failures to very few failures in the beginning, and the failures keep increasing with time).
Re:Official Google Blog: Sorting 1PB with MapReduce
by bd on Saturday November 22, 2008 @08:12PM
(#373354)
It's probable that they didn't use 48,000 brand new hard drives, but reused whatever they already had in their datacenters. Which would have a mixture of different drive ages, I guess.
Re:Official Google Blog: Sorting 1PB with MapReduce
by nebula on Sunday November 23, 2008 @02:46AM
(#373604)
That's an interesting possibility.
But the fact that the article specifically mentions about drive failures and that they use some of RAID to cope with this makes me wonder if they were using old disks. It's probably easier to get all new drives and use them isn't it? On second thoughts though, it might be much easier to tell GFS to make an extra mirror than to make sure that all disks are brand new. So, i buy your line of thinking :)
I'm a bit at lost here. Does that mean that one in ~50K hard disks fail within six hours of usage?