Skip to content

Comment on IBM builds 120 petabyte cluster out of 200,000 hard drives

Comments

With the current failure rate of harddisks, how many failing harddisks can one expect to swap out/in that array every day? I guess it must be a 24/7 job for a team of operators to keep that thing up. Or maybe they swap in spare harddisks by the cabinet...

At this kind of scale, you would probably just let disks fail which eventually would kill off entire logical modules, and simultaneously keep adding new modules to supplement them. So replacing the drives has no real urgency.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.