FW: What we need to log during experiments

["Douglas Olson" <dlolson@lbl.gov>]

-----Original Message-----
From: Henrik Nordberg [mailto:HNordberg@lbl.gov] 
Sent: Tuesday, September 29, 1998 10:47 AM
To: dlolson@lbl.gov
Cc: Arie Shoshani; Alexander Sim; Luis M. Bernardo; Doug Olson;
malon@anl.gov
Subject: Re: What we need to log during experiments


Doug Olson wrote:
> 
> We won't actually be running anything on linux.
> Doug

Hello,

Maybe we should. In the future we will use the Linux machines and thus
we have to move data over the network (from rmds03 to Linux). This will
take some time depending on the speed of link B (below). Since we are
moving quite a lot of data, this can be an important factor in the
overall system performance.

Can we run the user code on the Linux box this week? Even if that code
is "blind"?

 - Henrik
 
Background:
> > Some things I think we need to log when we run our experiments:
> >         ______                ________                 _______
> >        |      |       A      |        |       B       |       |
> >        | HPSS |--------------| RMDS03 |---------------| Linux |
> >        |______|              |________|               |_______|
> >
> >                                 CPU
<snip>
> > At runtime:
> > * A's transfer rate, both our part and other possible parts (we should
> > be alone) (CM + system (perfmon?))
> > * B's transfer rate (system)
> > * CPU usage on RMDS03 for our and other components (system)
> > * file size (CM)
> > * cache utilization (how full it is) (CM)
> >
> > Any more suggestions?
_________________________________________
Henrik Nordberg       <hnordberg@lbl.gov>
Scientific Data Management Research Group
Lawrence Berkeley National Laboratory