Don't the drive manufacturers bear some responsibility?

Story: The 2006 Linux File Systems WorkshopTotal Replies: 0
Author Content

Jul 14, 2006
10:52 AM EDT
While I like the idea of ZFS-like CRC checking of fs blocks on the fly (or "Self-Healing" if you are into Sun's Holistic/New Age terminology), I can't help but feel that the gloom and doom outlook regarding the future of hard drive reliablility is a bit far fetched.

Currently, IDE drives, and I assume SCSI too, detect bad blocks and remap them. They can run out of reserved blocks to use for remapping, and that is when you, or the FS start to notice a problem.

Sounds like in the future, the drive manufacturers are going to have to reserve more blocks.

Then again, the article also talks about errors that are not due to drive media failure: flying writes and such. These, if they are really a problem (I'd never heard of them before), also seem like something the hardware manufacturer needs to address. When one purchases a hard drive, he has the right to expect that it is more than a half-step above /dev/null with respect to reliability.

Which is not to say that adding checks in the FS and/or speeding up FSCK is a bad idea.

Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]

Becoming a member of LXer is easy and free. Join Us!