Re:SSD Failures in Datacenters
by bluecmd on Thursday July 07, 2016 @10:12AM
(#12049402)
I don't get where you got the notion that you need to understand the failure model to rely on it. It's certainly good to, and it will prevent over provisioning, but a lot of storage solutions today are so complex that understanding the failure models is practically impossible. S3? Super-black box.
Basically all you need to define is the expected output. "99.9pct of my reads are within 1ms" + "The read response is correct 99.9999% of the time" and then verify. Eject the disk if that doesn't hold, restore from replica. The rate of replacements is a factor of how close you are to the true failure model.
Basically all you need to define is the expected output. "99.9pct of my reads are within 1ms" + "The read response is correct 99.9999% of the time" and then verify. Eject the disk if that doesn't hold, restore from replica. The rate of replacements is a factor of how close you are to the true failure model.