Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!rpi!batcomputer!rogerj
From: rogerj@batcomputer.tn.cornell.edu (Roger Jagoda)
Newsgroups: comp.sys.next
Subject: Re: New performance monitor
Summary: Monitor Software
Keywords: Monitor
Message-ID: <9417@batcomputer.tn.cornell.edu>
Date: 11 Dec 89 23:15:11 GMT
References: <4289@hydra.gatech.EDU>
Reply-To: rogerj@tcgould.tn.cornell.edu (Roger Jagoda)
Followup-To: FQOJ@CORNELLA.CIT.CORNELL.EDU
Organization: Cornell Theory Center, Cornell University, Ithaca NY
Lines: 47

In article <4289@hydra.gatech.EDU> roy@lorien.gatech.edu (Roy Mongiovi) writes:
>A performance monitor (called, originally enough Monitor) is now available
>from the archives at cc.purdue.edu (and possibly others by now).  It can
>display up to 30 different statistics and combinations of statistics in a
>window, plus one statistic (or combination) in the icon, so it's kind of a
>combination of Sun's perfmon and perfmeter.
>
>  If anyone has any ideas how to improve this,
>I'd love to hear about it.
>
I think the program is super-fantastic. Really! I used it as the basis
of a report to justify a memory purchase for our servers (cubes also).
It's alsways easier to show management pictures. One problem though,
I had to use Scene to capture the windows and print them. Is there
room in you queue to add a "Print" button? I think the object is already
in the libraries as Terminal, Shell, and the others have it in their menus.
I'm not a programmer so I won't try and explain "it should be easy enough..."
type deals. If you have time, fine. Tell you what though, you could make this
shareware...we'd pay for it...:-)
 

>Please let me know if you find any problems with the statistics, or if you
>have improvements for the code or suggested enhancements.  One of the ideas
>I was toying with was to be able to point to a place on one of the graphs and
>have it display the actual value at that position in the graph numerically.
>Since that would take some effort to avoid having the mousedown handling code
>significantly degrade the updating of the display, and since I don't have the
>1.0 technical manuals yet, I decided to post before attempting that....
>
Well...as long as you asked...:-)
It would be nice to do "blow-ups." For example, maybe not in real-time
(to avoid the problem you mentioned) but how about pointing to a spot
on the graph and doing the equivalent of a "expand around this point"
scale expansion on just those stats so administrators could get a good
feel for more of what's happening. You could give the value of the 
variables tracked and their processes. This is REALLY useful when
diagnosing aberant "spikes" in I/O or page-outs. Just MHO!
 

>Roy J. Mongiovi     Systems Support Specialist     Office of Computing Services
>Georgia Institute of Technology	  Atlanta, Georgia  30332-0275   (404) 894-4660

Roger Jagoda
Cornell University
FQOJ@CORNELLA.CIT.CORNELL.EDU
NeXT Support Specialist