🚀 go-pugleaf

Two dimensional data redundancy is defined as RAID 6 by RAB.   StorageTek's
Iceberg uses this technology.  Unlike conventional RAID controllers that use
an XOR to compute redundant information, RAID 6 uses P+Q calculation - two
simultaneous equations with two unknowns.  Statistically this is much
greater
availability then a simple XOR protection scheme, however it does suffer
from
greater performance overhead (two disks must be updated on a partial stripe
write).

Roger
HP Enterprise Storage Division

Bill Todd wrote in message <7smbrs$7lb$1@pyrite.mv.net>...
>[Forgive the reply to myself - just to clarify:]
>
>I inadvertently munged two concepts together, I think:  I do understand how
>a 2-dimensional array arrangement works where there are m * n data disks
>(conceptually, an m by n rectangle) and m + n parity disks (arranged along
>two sides of the rectangle, providing parity in 2 dimensions), but think I
>have heard mention of some means of adding a second parity disk to a simple
>array that makes double-disk failures tolerable, which is what I'm curious
>about (since I don't see off-hand how it could be done).
>
>Thanks,  - bill
>
>Bill Todd <billtodd@foo.mv.com> wrote in message
>news:7sll6t$l8j$1@pyrite.mv.net...
>> But it still takes more disks:  3*n vs. 2*(n+1).  That said, given disk
>> prices these days, I wonder how important that is, even in read-mostly
(or
>> full-stripe-write) environments where RAID-5 has no performance penalty.
>>
>> I've never investigated the 'two-dimensional parity' approach that
>> supposedly uses 2 parity drives in a manner that allows *any two* drives
>in
>> a RAID set to fail simultaneously without data loss:  I assume it carries
>> greater write and rebuild overhead than RAID-5, but would be interested
in
>a
>> pointer to a good discussion.
>>
>> - bill
>>
>> Kjell Tore Nannestad <knannest@c2i.net> wrote in message
>> news:37EE4665.C2FF28E@c2i.net...
>> > I guess it's easier to make a second mirror using RAID 1.
>> > .
>> >
>> >
>> > prostorage wrote:
>> >
>> > > In article <37EB92A6.8579E99A@ultrafast.net.no.sp.am>,
>> > > RAIDER@ultrafast.net.no.sp.am says...
>> > > >Jim, let it be known that a hot spare drive in a RAID 5 setup does
>> > > >NOTHING to protect from a failure of more than one drive at a time.
>> What
>> > > >it does do, however, it minimizes MTTR (Mean Time To Repair) because
>> > > >most raid controllers will automatically initiate a rebuild of a
>failed
>> > > >drive onto a hot spare. Should another disk fail while the rebuild
is
>> > > >still in progress - the RAID set and all the data on it will be
lost.
>> > > >RAIDER
>> > > >
>> > > >Jim Pappe wrote:
>> > > >>
>> > > >> with Raid 5 plus hot spare - 2 drive can fail ..
>> > > >> www.raidexpert.com
>> > > >>
>> > >
>> > > RAIDER is correct
>> > > a second "warm" or "hot" spare would do nothing to save you if 2
>> > > drives went down at the same time
>> > >
>> > > the only way that i know of to live through the failure of 2 drives
>> > > simultaneously is to have 2 separate RAID 5's running as the
>> > > mirrored drives in a RAID 1 (& we've done this)
>> > >
>> > >   _____                         .     .
>> > >  '    \\                  .                .
|>>
>> > >      O//             .                        .                    |
>> > >     \_\          .                              .                  |
>> > >     | |      .   800-324-6711 / 800-720-7618 fax  .      .         |
>> > >    /  |  .       mailto:prostorage@earthlink.net   .   .    .      |
>> > >   /  .|   http://www.geocities.com/scsiperipherals/ . .      .     |
>> > >  / .  | http://kalypso.cybercom.net/scsiper/andys.html        .. . o
>> >
> ----------------------------------------------------------------------
>> > > Authorized - DIRECT VAR/VAD/Distributor for new SCSI/FC-AL
peripherals
>> > > from: IBM, Seagate, Quantum, Adaptec, Sony, Yamaha, ATTO, JNI, ATL
ect
>> >
>>
>>
>
>