(reiserfs) Re: More on 2.2.18pre2aa2 (summary of elevator ideas)

From: Rogier Wolff (R.E.Wolff@BitWizard.nl)
Date: Wed Sep 13 2000 - 10:25:31 EST


> "Jeff V. Merkey" wrote:
> >
> > One important point on remirroring I did not mention in my post. In
> > NetWare, remirroring scans the disk BACKWARDS (n....0) to prevent
> > artificial starvation while remirring is going on. This was another
> > optimization we learned the hard way by trying numerous approaches to
> > the problem.

Don't you want to do it a megabyte at a time to prevent abyssimal
disk-performance? (i.e. take the separate megabytes backwards, but do
every megabyte forwards)

instead of
        for (b=max;b>0;b--)
do
           for (bb=max-STEP;bb > 0 ; bb -= STEP)
                 for (b=bb;b<bb+STEP;b++)
(sloppy coding! this is pseudocde, don't copy-paste.)

I expect about 120 IO operations of say 4k (480k per sec) out of a
disk if you read strictly backwards. While you can get about 18 IOs of
1Mb per second (18M per second) out of a disk that you run a
(forwards) megabyte at a time.

The extra latency in the "elevator" is then the time to read a
megabyte, about 50ms, which sounds acceptable to me.

                                Roger.

-- 
** R.E.Wolff@BitWizard.nl ** http://www.BitWizard.nl/ ** +31-15-2137555 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
*       Common sense is the collection of                                *
******  prejudices acquired by age eighteen.   -- Albert Einstein ********

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Fri Sep 15 2000 - 21:00:21 EST