User: Password:
|
|
Subscribe / Log in / New account

Infrastructure unification in the block layer

Infrastructure unification in the block layer

Posted Oct 21, 2009 13:03 UTC (Wed) by job (guest, #670)
Parent article: Infrastructure unification in the block layer

Many thanks for an interesting article.

One thing that was not clear to me is what vendors the "vendor specific" metadata of MD refers to. Probably not Linux distribution vendors.

What would be interested is if someone with a knowledge of GEOM could highlight some of the differences and similarities of that scheme. It is the latecomer of the bunch and is now used all over the place in FreeBSD, at least until ZFS came and shook up block device management again.


(Log in to post comments)

Infrastructure unification in the block layer

Posted Oct 21, 2009 17:11 UTC (Wed) by dlang (subscriber, #313) [Link]

I believe (but could be wrong) that vendor specific metadata is referring to raid metadata that commercial raid vendors put on drives. MD is able to understand that data and assemble arrays based on it (and avoid overwriting it)

Infrastructure unification in the block layer

Posted Oct 22, 2009 0:09 UTC (Thu) by nix (subscriber, #2304) [Link]

This is *damn* good because it means that if your proprietary RAID card
dies, md can get the data off the disks again. (It can probably do it
anyway with RAID-5 as long as you remember the stripe size, as there
aren't many ways to arrange a RAID-5 array. But RAID-6 is harder.)


Copyright © 2017, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds