Software RAID issues (was Re: Suggestions solicited, server bring up)

Ben Scott dragonhawk at gmail.com
Tue Nov 24 14:13:27 EST 2009


On Mon, Nov 23, 2009 at 9:25 AM, Tom Buskey <tom at buskey.name> wrote:
> RAID5 will have faster read performance then RAID 1 or a single disk.

  This is possible if there are multiple I/O paths.  Many controllers
don't do that; there is a single I/O path to the RAID engine.  So
adding disks actually slows things down.

  As has been suggested, implementation details matter a lot here.

> I think the only other
> filesystems that checksum are NetApp's WAFL(?) and Linux's btfrs.

  In an interesting blast from the past, MS-DOS FAT maintains a
checksum.  IIRC, it's 32-bit CRC across the entire file, so it isn't
very robust, but I've seen it successfully detect errors before.  So
MS-DOS FAT actually has a feature Linux EXT3 lacks.  Whodathunkit?

> When Solaris does an upgrade, it creates a new partition for the
> upgrade OS alongside the original.

  You can do this on Linux with LVM snapshots.

-- Ben


More information about the gnhlug-discuss mailing list