Raid 5 performance

[Lennart Sorensen]

On Tue, Sep 21, 2004 at 05:42:16PM -0400, Ralph Doncaster wrote:
> To avoid having to read from the odd data disk in order to calculate the
> checksum when doing a write.

You have to know the value of n-1 disks in order to calculate the xor
value for the last disk, so you either have to read before write on one
disk, or read from all disks and do a write.  I don't see how either is
affected by number of disks.

If you read the xisting checksum and existing value of the drive you are
writing a block to, you know all that you need to know to recalculate
the checksum, no matter how many disks are involved in that checksum.
Reading all the disk and calculating the checksum from scratch is a good
way to verify the data integrety as well, so I suspect that is what is
usually done.

Lennart Sorensen
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml