[KLUG Members] Re: RAID

[Bryan J. Smith]

--=-S614M4sx9Sjra2enAtcL
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On Sun, 2002-12-15 at 13:24, Peter Buxton wrote:
> No, I think he meant "no matter what you use, make sure you have a
> couple spares handy," which is obviously more important (safety) than
> performance.

Well, that too -- and I gathered that from the rest of his post.  But I
wanted to point out the same thing that is available in cheap PCI cards
are basically what you have on the mainboard.  If you have a Promise
controller on your mainboard, you should be able to read its volumes
with an equivalent series Promise "FastTrak" card.

Furthermore, I also thought he was striving for the fact that one should
stay away from the popular ATA RAID approach, especially with his
preference for IBM i960-based SCSI RAID products.  There is a _huge_
difference in the design of "BIOS-less" ATA RAID and card that has an
on-board, intelligent microcontroller, an embedded OS, DRAM buffer and
4+ ATA channels proliferating the use of only _one_ device per channel.

> Frankly, I didn't buy this board for the RAID chip. I wasn't at all
> interested in RAID, and I had only two partitions on one drive anyway:
> swap and /, until I recently became interested in multidisks.

Understand that "regular" ATA =3D "BIOS-only" ATA RAID -- the _hardware_
is 100% _the_same_!

It wasn't long after off-chipset ATA became popular that the ATA
controller designers realized they could add a little RAID volume
organization to their on-board, 16-bit BIOSes and then license some ATA
RAID code from a 3rd party for use in the 32-bit OS driver.  It's a very
cheap way to offer RAID -- virtually no cost over none.  Who cares if it
doesn't scale, most people want cheap, not fast or scalable.

> I'm also perfectly happy using ATA and not SCSI (I'm such a Philistine :)=
.

For multiple ATA disks, ATA can become a burden on the I/O and CPU.=20
Especially if you throw CPU-driven, software RAID (regardless whether or
not it is those "cheap" cards or OS-based RAID) into the mix.  And
never, _ever_ put more than one ATA drive on an ATA channel if you are
accessing both of them.

SerialATA doesn't even let you do that -- one device per channel
(finally solving that long-standing performance issue with ATA).=20
SerialATA finally addresses a lot of things wrong with ATA.

Of course it still doesn't solve some things with any ATA in general.=20
The lack of command queuing, the "dumb" block I/O, the lack of
device-to-device transfers.

And while ATA-designed controllers like the 3Ware Escalade products
address some of these, they still don't address them all.  People will
continue to choose SCSI over ATA in those situations, just as FireWire
is a better solution than even USB 2.0 for similar reasons.

> In any case, I'll take the advice of the kernel hackers, which I had not
> read carefully enough the first time:
> CONFIG_BLK_DEV_ATARAID
>   Say Y or M if you have an IDE Raid controller and want linux
>   to use its softwareraid feature.  You must also select an
>   appropriate for your board low-level driver below.
>   Note, that Linux does not use the Raid implementation in BIOS, and
>   the main purpose for this feature is to retain compatibility and
>   data integrity with other OS-es, using the same disk array. Linux
>   has its own Raid drivers,=20
>   *which you should use if you need better performance.*

Yes, I made all those points in my previous E-mail.

The GPL Linux driver basically replaces the licensed, 3rd party ATA RAID
implementation why the kernel developers don't have access to.  That
code is only available in binary-only form via the binary-only Windows
or Linux (kernel version-fixed) drivers.  You couldn't use the BIOS RAID
anyway because it is 16-bit!  Not Windows nor Linux can use it.


--=20
Bryan J. Smith, E.I. (BSECE)       Contact Info:  http://thebs.org
[ http://thebs.org/files/resume/BryanJonSmith_certifications.pdf ]
------------------------------------------------------------------
  The more government chooses for you, the less freedom you have.


--=-S614M4sx9Sjra2enAtcL
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQA9/PinDjEszaVrzmQRAtwBAKC7pTgt6QvtDRydF0/7EMNlCOC9zACffysf
RilGUHzJlD3VIDdf1sog+2k=
=Opzk
-----END PGP SIGNATURE-----

--=-S614M4sx9Sjra2enAtcL--