lkml.org 
[lkml]   [2005]   [Jun]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [RFC] blkstat
Date
On Mon, 13 Jun 2005 20:02, Nick Piggin wrote:
> Con Kolivas wrote:
> > On Mon, 13 Jun 2005 19:46, Nick Piggin wrote:
> >>Oh, and before I go further, does anyone know of any program
> >>or statistic that allows the same functionality? Any comments?
> >
> > Would something like iostat give similar results?
>
> The problem with that is that it does not give you a % idle
> figure on the block device, so you basically can't see if
> the device is becoming a bottleneck.

I've often wondered how iostat gives a %busy figure and whether this
translated accurately without further info from the kernel.

> You can kind of guess if you take into account the seeks,
> and the throughput, but you're still missing things like
> head position (eg. changes throughput), settle time and
> rotational latency, and lots of other stuff.
>
> Thanks,
> Nick
>
> Also, BTW. the way I have done the kernel patch make a
> device show 100% utilisation even if it is not doing anything
> but waiting for a plug, or an anticipatory scheduler. This
> is basically all the end user wants to know, although for
> development purposes it may be interesting to know the other
> metric too.

That sounds quite useful.

Cheers,
Con
[unhandled content-type:application/pgp-signature]
\
 
 \ /
  Last update: 2005-06-13 12:10    [W:0.080 / U:0.064 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site