Re: The reason to call it 3.0 is the desktop (was Re: [OT] 2.6 not 3.0 - (NUMA))

From: Simon Kirby (sim@netnation.com)
Date: Mon Oct 07 2002 - 21:36:54 EST


On Mon, Oct 07, 2002 at 12:36:48PM -0700, Andrew Morton wrote:

> Block allocators are fertile grounds for academic papers. It's
> complex. There is a risk that you can do something which is
> cool in testing, but ends up exploding horridly after a year's
> use. By which time we have ten million deployed systems running like
> dogs, damn all we can do about it.
>
> The best solution is to use first-fit and online defrag to fix the
> long-term fragmentation. It really is. There has been no appreciable
> progress on this.
>
> A *practical* solution is to keep a spare partition empty and do
> a `cp -a' from one partition onto another once per week and
> swizzle the mountpoints. Because the big copy will unfragment
> everything.

Having seen fragmentation issues build up on (mbox) mail spools over
several years first hand, I can say that mail spools definitely show the
need for a defragmentation tool. I remember actually doing the "cp -a"
trick just to restore the mail server to decent performance (which
worked amazingly well, for another few months). (This was before we
switched to hashed directories and a POP3 server which caches mbox
messages offsets/UIDLs/states.)

Being able to defragment online would be very useful. I've seen some
people talk about this every so often. How far away is it?

Simon-

[ Simon Kirby ][ Network Operations ]
[ sim@netnation.com ][ NetNation Communications ]
[ Opinions expressed are not necessarily those of my employer. ]
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon Oct 07 2002 - 22:01:03 EST