Re: RAID 5 q - in simpler terms - THANKS!
From: Leyden, Joseph (LeydenJ_at_METRO.NET)
Date: 02/16/05
- Previous message: Bruce Whittaker: "Re: RAID 5 q - in simpler terms - THANKS!"
- Maybe in reply to: Leyden, Joseph: "Re: RAID 5 q - in simpler terms - THANKS!"
- Next in thread: Roger Deschner: "Re: RAID 5 q - in simpler terms - THANKS!"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 15 Feb 2005 15:35:53 -0800 To: aix-l@Princeton.EDU
I used to work for the R&D division of Citibank (the 1st in ATM) and the
only requirement that they have adopted for any design is -> NO SINGLE POINT
OF FAILURE. A mirrored rootvg is sufficient. Of course a mirrored off-site
rootvg is better if you think the a TSUNAMI or Earthquake will collapsed the
entire building. After a certain point, PESSIMISM could be a misuse of your
imagination.
Joe
-----Original Message-----
From: Bruce Whittaker [mailto:bwhittak@ENERGY.COM.AU]
Sent: Tuesday, February 15, 2005 2:42 PM
To: aix-l@Princeton.EDU
Subject: Re: RAID 5 q - in simpler terms - THANKS!
Hey - what's this years past stuff ??
We're still running a Raid5 using 5x4Gig drives - and it's been so solid I'm
hesitant to ever go off Raid-5.
Of course it's now getting really really difficult to replace a 4 gig drives
because they are becoming very hard to find.
By the way - I used to work in a nuclear environment - a double failure is
expected, so everything was triplicated.
I admit - it's time to migrate now....
Bruce Whittaker,
----------------------------------------------------------------------------
------------
This e-mail may contain confidential or privileged information. If you
have received it in error, please notify the sender immediately via return
e-mail and then delete the original e-mail. EnergyAustralia has collected
your business contact details for dealing with you in your business
capacity. More information about how we handle your personal information,
including your right of access is contained at http://www.energy.com.au.
----------------------------------------------------------------------------
------------
BRUCE HARVEY <btharvey@MANDTBANK.COM>
Sent by: IBM AIX Discussion List <aix-l@Princeton.EDU>
16/02/2005 05:41
Please respond to IBM AIX Discussion List
To: aix-l@Princeton.EDU
cc:
Subject: Re: RAID 5 q - in simpler terms - THANKS!
... Never say "never" ... We thought that a triple failure would never
happen. We discovered that it would "seldom" happen. Even IBM drives fail,
and especially drives that have been spinning for years ... if they never
spin down, sometimes they forget how to spin up. ;-) Just think of those 4
Gb drives of years past (got your hammer?).
Bruce
>>> LeydenJ@METRO.NET 2/15/2005 12:57:15 PM >>>
TRUE! primary and secondary mirrored drive failing at the same
time is an exception to a 'downtime' but in our experience with
IBM drives or IBM recommended drives this has never happened.
But you have a point there on whose make of drives you use.
Joe
-----Original Message-----
From: Hunter, Mark [ <mailto:Mark.Hunter@ANHEUSER-BUSCH.COM]>
mailto:Mark.Hunter@ANHEUSER-BUSCH.COM]
Sent: Tuesday, February 15, 2005 9:31 AM
To: aix-l@Princeton.EDU
Subject: Re: RAID 5 q - in simpler terms - THANKS!
On SSA, we use 0+1, all software based. Using mklv with -u, -S, and -c.
This
tends to balance I/O better than we can by hand. The complexity increases
because certain commands change behavior if it's a striped logical volume.
For
example, if its striped on disks 1-4, then all expansion will be on disks
1-4.
Also, if you lose one disk in a stripe, the sync requires all disks in the
stipe
to be resynced. The -P option to syncvg is your friend.
We are in the process of moving to hardware based Raid-5 for almost all of
our
systems for non rootvg data. We have an enterprise Hitachi SAN. The
economy of
scale helps the cost plus it saves a lot of floor space.
On your statement that no downtime is ever involved on mirroring, that's not
strictly true. If the wrong two drives fail, you will have downtime. We
have
had situations where both the primary and secondary copies of a mirror have
failed at nearly the same time. 4 drives mirrored to 4 other drives, but
the
"wrong" two drives failed. With raid-5, its any two drives, so it's a
greater
chance, but with autospare we have not seen it happen.
We have found that the Hitachi Raid-5 SAN solution is at least as reliable
as
most mirroring options in practice.
Mark Hunter
Anheuser-Bus ch Cos.
MIS Consultant, ES&SO Server Planning and Integration
*Office: (314) 632-6663
*Fax: (314) 632-6901
*Pager: (314) 841-4026
*Email: Mark.Hunter@Anheuser-Busch.com
The information transmitted (including attachments) is covered by the
Electronic
Communications Privacy Act, 18 U.S.C. 2510-2521, is intended only for the
person(s) or entity/entities to which it is addressed and may contain
confidential and/or privileged material. Any review, retransmission,
dissemination or other use of, or taking of any action in reliance upon,
this
information by persons or entities other than the intended recipient(s) is
prohibited. If you received t his in error, please contact the sender and
delete
the material from any computer.
-----Original Message-----
From: IBM AIX Discussion List [ <mailto:aix-l@Princeton.EDU]>
mailto:aix-l@Princeton.EDU] On Behalf Of
Green,
Simon
Sent: Tuesday, February 15, 2005 4:52 AM
To: aix-l@Princeton.EDU
Subject: Re: RAID 5 q - in simpler terms - THANKS!
That's generally my view. Of course, when you start talking about
multi-terabyte arrays the potential cost savings are considerable. But for
smaller stuff, if you're just using SSA or something like that RAID-0 is
much
simpler to manage and gives excellent results.
Some people favour RAID 0+1, where you stripe the data - simple RAID 1 with
no
parity - then mirror it, (or vice versa: it can make a difference). This
gives
you something of the best of both worlds but it adds to the complexity.
-- Simon Green Altria ITSC E urope s.a.r.l. AIX-L Archive at <https://new-lists.princeton.edu/listserv/aix-l.html> https://new-lists.princeton.edu/listserv/aix-l.html New to AIX? <http://publib-b.boulder.ibm.com/redbooks.nsf/portals/UNIX> http://publib-b.boulder.ibm.com/redbooks.nsf/portals/UNIX N.B. Unsolicited email from vendors will not be appreciated. Please post all follow-ups to the list. > -----Original Message----- > From: IBM AIX Discussion List [ <mailto:aix-l@Princeton.EDU]> mailto:aix-l@Princeton.EDU] On Behalf > Of Leyden, Joseph > Sent: 14 February 2005 18:56 > To: aix-l@Princeton.EDU > Subject: Re: RAID 5 q - in simpler terms - THANKS! > > > Thanks! You've answered my question with the overhead being the parity > data. I mentioned the word 'compressed' so we can talk strictly in > terms of bytes. > > Therefore (on the array of 5 disk say 1 disk is 9 GIG) I can not have > 45 gig of application data but only about 32 GIG. AND some time > overhead in rebuilding the replaced failed disk. > > I'm inclined to think that if I can afford disk (1 to 1) mirroring I > should stick with it since NO downtime is ever involved.
- Previous message: Bruce Whittaker: "Re: RAID 5 q - in simpler terms - THANKS!"
- Maybe in reply to: Leyden, Joseph: "Re: RAID 5 q - in simpler terms - THANKS!"
- Next in thread: Roger Deschner: "Re: RAID 5 q - in simpler terms - THANKS!"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
