On Wed, Jun 26, 2002 at 02:43:46PM -0400, Steven M. Bellovin wrote:

> >That's too bad.  It would be nice if it could break order after a
> >process had passed some I/O limit.  It limits the effect of a common
> >tuning method whereby you move applications to separate drives.
> 
> Agreed.  I've upgraded several of my machines to 1.6beta2 and beta3, 
> and I'm seeing the same symptoms that Wojciech has mentioned.

I am hoping that in the long run UBC is going to be a big win though.
My old hardware works fine, and I'm way too lazy to replace it.  I know
what will happen if I get faster hardware: new project... :)

> Unix-like systems have long had this problem -- the disk-scheduler is 
> too far removed from any process's context.

Well, that's largely been on purpose.  It can result in pretty good
overall throughput.  I don't think it would be too hard to link processes
more closely.  The first shot might be rough, like first-run SMP kernels,
but maybe a research fork could explore some ideas.

This is the sort of thing which would seem ideal for company sponsored
research.


-- 
UNIX/Perl/C/Pizza__________________________________shannon@widomaker.com