[ILUG] Highpoint Sata RAID
Paul Jakma
paul at clubi.ie
Wed Mar 23 19:27:02 GMT 2005
On Wed, 23 Mar 2005, FRLinux wrote:
> On Wed, 23 Mar 2005 09:24:33 -0800, Rick Moen <rick at linuxmafia.com> wrote:
>> And, no, the HighPoint 1820A and Promise SATA150 TX4 cards don't have
>> the "same performance" as real hardware RAID: Only the XOR logic is
>> handled by a dedicated chip. All other RAID logic is offloaded to the
>> system CPU and OS.
Ah, apparently Rick replied to me with:
"Quoting Paul Jakma (paul at clubi.ie):
> If that performance sensitive stuff is on
> the card, who cares whether management is done on a card CPU or a
> host CPU?
People who care about avoiding crummy manufacturer-proprietary disk
schemes care. People who resent having to buy the second card again
to get their data back if/when the first one fails care."
I have Rick kill-filed (for reasons consistent with above replies).
Let's be clear:
- The Promise TX4 is *not* the same card as the Promise SX4[1]. (it
doesnt have the offload features). Yet Rick is replying to some
imaginary thread where we're discussing the TX4.
- If you're using Linux MD, as you should be, (be it with a promise
TX4 or SX4 or Highpoint or whatever) then you're
obviously not using proprietary on-disk RAID formats..
- On-disk RAID superblock format is *not* the cause of performance
problems with soft-raid.
- The primary reason soft-raid can be slower is because it requires
extra bus bandwidth - instead of writing a block once and letting
the controller write a copy to each disk you have to send each
block across the bus for each disk (for RAID-1). For RAID-5 the XOR
doesnt cost much on a modern CPU (which are *vastly* superior in
speed and RAM bandwidth to the CPUs and RAM used on "embedded
computer on a PCI card" RAID controllers), however it is an extra
block of data to pass across the bus.
Ie, extra bus bandwidth of software RAID is the primary bottleneck
of soft RAID
- The Promise SX4 has onboard RAM and XOR, which the OS potentially
can use to eliminate the aforementioned #1 soft-raid performance
limitation.
- Linux MD doesnt support such offload yet, however hopefully it will
at some stage in the feature, and if you have an SX4 that'll be a
nice free performance boost when it does.
However, please note that 500 euro+ SCSI hardware RAID cards +
expensive and low GB/euro SCSI disks are vastly more reliable than
and superior to cheap S-ATA cards and high GB/euro (yet still cheap)
S-ATA disks. I would urge all my competitors to go with the SCSI +
hardware RAID option...
1. Much the same way as parallel ATA is not quite the same thing as
serial ATA...
regards,
--
Paul Jakma paul at clubi.ie paul at jakma.org Key ID: 64A2FF6A
Fortune:
You will gain money by an immoral action.
More information about the ILUG
mailing list