RE: Quick Survey

Joe
 a three drive  RAID set is not my idea of good.
FOUR minimum.

At 03:28 PM 9/2/2005, Joe Pluta wrote:

> From: Chris Bipes
>
> How critical is not having down time?
>
> Raid 1 - full mirror - Can loose multiple drives in and still be
running
> if you do not loose both drive that are mirrored together.
>
> Raid 5 - 3+ drives utilizing % of each drive to stripe data from other
> drives.  Loosing more than one in a raid set causes loss of all data
in
> that set.

Seems to me the exposure on these are very close to equal.  The chances
of losing two drives is miniscule, and there's just about as much chance
of losing two in the same RAID1 set as losing two anywhere in a RAID5
set (I know it's not mathematically the same, but from an order of
magnitude position, you can kill a RAID1 set almost as easily as a RAID5
set).


> Mirrored Raid 5 - The best of both worlds and your data is most
> available.  Still loosing multiple drive in each of the mirrored Raid
5
> sets will cause you to loose all the data in that Raid 5 array..

This is probably a reasonable option for mission critical H/A systems.
RAID50 is the fastest and most redundant system.  However, it requires
quite a bit of extra disk; as much as 200% overhead worst case.


> Mirrored system - complete redundancy of server.  Should also have
> redundant backplanes in redundant building in redundant locations.
>
> Why depends on how critical the application is and amount of down time
> and/or data loss management will tolerate.

And redundant Internet connections and everything else.  Different
league.  Again, that's why I was specifically talking about RAID1 vs.
RAID5.  These seem to be the reasonable choices for anything short of a
realtime mission critical system.

Of course, the next discussion is hot-swap vs. non-hot-swap <g>.  Again,
for a non-mission-critical system where loss of work is more important
than interrupted service, it seems that hot-swap is an unnecessary
expense.  On the other hand, the ability to quickly replace a failing
drive before a second one goes casters up is probably a good reason to
have hot-swap bays in HA environments.

Joe

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.