SUMMARY: OSF/1 3.0 5 minute performance spikes

From: Saul Tannenbaum <stannenb_at_emerald.tufts.edu>
Date: Wed, 1 Mar 1995 23:29:19 -0500 (EST)

Subject: SUMMARY: OSF/1 3.0 5 minute performance spikes

I posted a query regarding once-every-5-minute, across the board
performance spikes we've been seeing on a heavily loaded OSF 3.0
system.

We've isolated the problem to the Polycenter Performance Solution
data archiving function. When you archive data, you see these
spikes, which, in turn, drive the load average to the roof. With
no archiving, you get no spikes.

It's not at all clear to me what good a Performance monitor is that
causes such significant artifacts in the data it collects, and one
that inhibits performance greatly.

Thanks to those who answered me:

        Mike Iglesias <iglesias_at_draco.acs.uci.edu>
        Arrigo Triulzi <a.triulzi_at_ic.ac.uk>
        Jim Wright <jwright_at_phy.ucsf.edu>
        Richard L Jackson Jr <rjackson_at_gmu.edu>
        eric_at_diablo.slc.ca.gov
        jnolan_at_ibm.net
        Stuart Kreitman <skk_at_benelux.pa.dec.com>

        - Saul

-- 
Saul Tannenbaum, Manager, Academic Systems | "It's still rocket  
                stannenb_at_emerald.tufts.edu |    science" - Vint Cerf
Tufts University Computing and             |
                Communications Services    |

Received on Wed Mar 01 1995 - 23:30:03 NZDT