Query on kernel metrics available

Query on kernel metrics available

[Raj Mathur]

A competing OS vendor is claiming that the Linux kernel has no way to 
measure the following:

a. Time spent waiting for page faults in percent

b. Total transfer time for each auxiliary memory in percent

c. Time in percent when one or more tasks were blocked waiting for 
memory resources

d. Time spent waiting for page faults in percent

I was wondering if the claims were pure FUD or are measurements 
possible -- whether they are actually implemented or not is not 
important (at least at this stage).  Apologies if the descriptions 
aren't very precise; OTOH that leaves us room to interpret the claims 
in a favourable (to us) manner :)

TIA and regards,

-- Raju
-- 
Raj Mathur                raju@kandalaya.org      http://kandalaya.org/
 Freedom in Technology & Software || February 2008 || http://freed.in/
       GPG: 78D4 FC67 367F 40E2 0DD5  0FEF C968 D0EF CC68 D17F
PsyTrance & Chill: http://schizoid.in/   ||   It is the mind that moves

Re: Query on kernel metrics available

[Alan Cox]

On Mon, 19 Nov 2007 12:59:26 +0530
Raj Mathur <raju@linux-delhi.org> wrote:

> A competing OS vendor is claiming that the Linux kernel has no way to 
> measure the following:
> 
> a. Time spent waiting for page faults in percent
> 
> b. Total transfer time for each auxiliary memory in percent
> 
> c. Time in percent when one or more tasks were blocked waiting for 
> memory resources
> 
> d. Time spent waiting for page faults in percent

a an d appear to be the same thing

> I was wondering if the claims were pure FUD or are measurements 
> possible -- whether they are actually implemented or not is not 
> important (at least at this stage).  Apologies if the descriptions 
> aren't very precise; OTOH that leaves us room to interpret the claims 
> in a favourable (to us) manner :)

Systemtap as shipped by most of the enterprise vendors can measure all of
this quite easily. 

Alan