Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!magnus.acs.ohio-state.edu!csn!hellgate.utah.edu!cc.usu.edu!jrd
From: jrd@cc.usu.edu (Joe R. Doupnik)
Newsgroups: comp.dcom.lans
Subject: Re: Rate of change vs thresholds.
Message-ID: <1991Jun26.163245.48241@cc.usu.edu>
Date: 26 Jun 91 22:32:45 GMT
References: <1991Jun26.190512.3937@shl.com>
Organization: Utah State University
Lines: 28

In article <1991Jun26.190512.3937@shl.com>, phil@shl.com (Phil Trubey) writes:
> Here's a question for you network management gurus out there...
> 
> I'm wondering if there are any examples where you would want to monitor
> the rate of change of some network variable rather than the absolute
> value.  Eg. Typically people (or their network management stations)
> monitor link utilization levels and want to be notified if the 
> utilization rate ever hits a certain level (eg 60% or whatever
> for an Ethernet network).  Have people found that there
> are certain variables that they wished they could monitor the
> rate of change of that variable (eg, how quickly the Ethernet
> utilization level is changing)?
> 
> I'm just wondering if this is a useful feature or not.
> 
> Phil Trubey                     |  Internet: phil@shl.com  
> SHL Systemhouse Inc.            |        or: phil%shl.com@uunet.uu.net
> 50 O'Connor St., Suite 501      |  UUCP:     ...!uunet!shl!phil
> Ottawa, Ontario, Canada         |  Phone:    613-236-6604 x667
> K1P 6L2                         |  Fax:      613-236-2043
-----------------------------------
Phil,
	Yes, rates of change can be an important parameter: forstalling
disaster. But a program will have to do some fancy filtering to extract
meaningful results. The reason is, obviously, network activity is highly
irregular and not well described by simple random variable distribution
functions.
	Joe D.