Re: Disk Mangement Advice

From: Phillip Windell (_at_.)
Date: 01/06/05


Date: Thu, 6 Jan 2005 12:17:12 -0600


"Phillip Renouf" <PhillipRenouf@discussions.microsoft.com> wrote in message
news:2DD332DE-C485-496F-A186-B6A0C0F87AB3@microsoft.com...
> Writing to a RAID5 set is a much bigger performance hit than writing to a
> RAID1 set. In a RAID1 set you are only writing to 2 drives, but in a RAID5
> set you are writing to 3+ drives and it also has to calculate the parity
> information.
>
> Also, the n-1 does not denote that one drive holds the parity data:
> effectively with a RAID5 set the available drive space is effectively
equal
> to n-1. The parity data is actually spread across all the drives.

No, the fact that it is 3 drives (or 4 or 5) doesn't not make it slower. In
fact it is faster because it writes to all of them at same time and does not
have to write the data twice. The data is also evenly divided across the
drives so that only a portion of the data need to be writen to each one.
There is a slight time loss for the Parity but the "cost" of this negligable
and often exaggerated and is nearly compensated for by the faster
writetimes. There is also two RAID5's,...there is RAID5 with Parity and
Raid5 without Parity (Stripped Set). The RAID5 without Parity is the fastest
RAID of them all.

RAID1 is slower because and entire copy of the data goes to the second drive
so the data being completely written twice in its entirety creates a "cost".

-- 
Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com


Relevant Pages

  • Re: Disk Mangement Advice
    ... > effectively with a RAID5 set the available drive space is effectively ... The parity data is actually spread across all the drives. ...
    (microsoft.public.windows.server.networking)
  • Re: Disk Mangement Advice
    ... > effectively with a RAID5 set the available drive space is effectively ... The parity data is actually spread across all the drives. ...
    (microsoft.public.windows.server.setup)
  • Re: RAID question
    ... |> on the parity block of a stripe in a RAID5 setup is generated by XORing ... I'll have to read up on RAID5 implementation. ... or is XOR chosen as implementation in some/all designs? ... If you are getting data from all drives ...
    (comp.os.linux.misc)
  • Re: RAID question
    ... |> on the parity block of a stripe in a RAID5 setup is generated by XORing ... I'll have to read up on RAID5 implementation. ... or is XOR chosen as implementation in some/all designs? ... If you are getting data from all drives ...
    (alt.os.linux)
  • LSM raid5 volume causing errors
    ... I was just working on an LSM raid5 issue for which I posted my summary just ... The way I have that test bed set up is with a couple of 4gig drives for the ... raid5 set, all that on just one bus on a KZPBA-CX controller. ... I have the log plex on the drive 0, which is one of the two disks from the ...
    (Tru64-UNIX-Managers)