Coda File System

Re: venus/kernel profiling/speed

From: Troy Benjegerdes <hozer_at_hozed.org>
Date: Fri, 16 Jul 2004 10:58:22 -0500
On Fri, Jul 16, 2004 at 08:08:53AM +0200, Ivan Popov wrote:
> On Thu, Jul 15, 2004 at 10:53:39AM -0500, Troy Benjegerdes wrote:
> > I suppose part of it is data coming from kernel -> venus -> kernel ->
> > disk. Can we find a way to short-circuit the common case so the bulk
> > file data doesn't have to go through copy_{to|from}_user in the kernel?
> 
> Bulk file data does not hit venus, kernel is doing disk i/o itself,
> venus is seeing only open/close. So performance difference against a
> local file system can be only for directory operations, which it
> probably is in your test case - creation of lots of files.
> 

It's *really* *really* slow at handling file opens/creates/closes then.
Untarring a linux kernel seemed to take over half an hour, and during
portions of that time venus was taking over 50% of the cpu, while bzip2
was only getting about 30%.

The other thing that seems a heck of a lot slower than it should be is
the checkpoint.. I am using the debian packages (rebuilt on ppc).. My
first thought was it was build non-optimized, but that doesn't seem to
be the case.

-- 
--------------------------------------------------------------------------
Troy Benjegerdes                'da hozer'                hozer_at_hozed.org  

Somone asked my why I work on this free (http://www.fsf.org/philosophy/)
software stuff and not get a real job. Charles Shultz had the best answer:

"Why do musicians compose symphonies and poets write poems? They do it
because life wouldn't have any meaning for them if they didn't. That's why
I draw cartoons. It's my life." -- Charles Shultz
Received on 2004-07-16 12:00:44