At 16:04 13-11-2001 +0100, Roy Sigurd Karlsbakk wrote:
> The fs blocksize has not much to do with the amount it writes in one go.
> XFS uses delayed allocation so it will probably write a lot of data at once
> anyways.
How about reading? The reading is really essensial here - writing is done
by batch, and can take whatever time it likes (almost :-)
That should be even better. I noted Higher read speeds with 2.4.14 -linus
but worse write speeds. It sounds like it has what you are looking for.
> If you can give some
que?
Ehm...yes. This was the part where I was about to ask about your hardware
availability and setup. But I got interrupted and then sent out the email.
(I'm like a goldfish. My short term memory spans about 3 seconds ;-)
You said in the other mail that you got 30MB/s. That is not right. I
already reach about 70MB/sec on a 6 disk 10K RPM raid 10 on hardware raid.
Software raid is even faster then hardware raid most of the time.
The newer ami megaraid firmware finally does read balancing on all disks
instead of one of the raid set drives.
A IDE disk already has 30MB/s sequential read so something is really "in
the way". You have the disks on multiple scsi channels over multiple cards
which are 64 bit PCI cards ofcourse. :-)
What motherboard do you use?
Cheers
--
Seth
Every program has two purposes one for which
it was written and another for which it wasn't
I use the last kind.
|