Prediction

[mark]

I looked at Cricket and it does prediction, but not forecasting (AFAIK). 
What Cricket does, to my understanding, is look for times when the metric 
is outside the baseline bounds. These baseline bounds are determined from 
past history. So Cricket watches the current metric value and if it is 
outside the bounds, it flags it for further investigation. This reduces 
the need for humans to have to watch graphs on a regular basis. They only 
need look at the graphs and times that have been flagged. You can achieve 
similar results with thesholds but you still need someone to set the 
threshold for each metric. Using Cricket, I think you can automatically 
get baseline or threshold values over time.

What I would like is forecasting or prediction into the future. Basically, 
if a certain rate of change has been logged over some historical time 
period, project that out into the future so I can see when we will run 
out of disk, cpu, memory, whatever. 

But thanks for the input. I'm also looking at using Cricket to reduce the 
human labor needed to watch things and the human "knowledge" needed to 
correctly set thresholds.

Mark

On Fri, 18 Apr 2003, Jamie 
Baddeley wrote:

> Hi Mark,
> 
> I think this has been done. The stuff that Mr Hopcroft referred to 
> (Holt-Winters forecasting) has been implemented in Cricket (which came out of 
> webtv)
> 
> http://cricket.sourceforge.net/aberrant/rrd_hw.htm
> 
> I must admit I haven't used it.
> 
> jamie
> 
> On Fri, 18 Apr 2003 00:58, mark wrote:
> > I think that would be a useful feature. I think howerver instead of just
> > "prediction" we should talk more about growth and capacity planning.
> > Prediction may be confused with failure prediction. With the recent
> > discussions around APAN, Orca, rrd, etc, etc. I can see that prediction
> > could be implemented in the performance graphing arena. For example, if
> > you are graphing the historical trend of a metric, it should be reasonable
> > that you could extrapolate that line out into the future. Of course the
> > more history you have, the more likely that projection is to be accurate.
> >
> > Interesting idea. Jim, I know you've been looking at jgraph, is it capable
> > of extrapolation? If not, does anyone else know of a plotting package that
> > is? Maybe phplot or gnuplot?
> >
> > Mark
> >
> > On Thu, 17 Apr 2003, Steve Freegard wrote:
> > > Hello,
> > >
> > > I've had an idea for a useful plug-in: prediction - e.g. Based on current
> > > hourly/daily/weekly/monthly usage, how many hours/days/months/years will
> > > it take for a device (e.g. disk/avg.cpu/avg. memory/avg. bandwith) to run
> > > out of capacity or reach a set value (e.g. 80% avg cpu usage), and to be
> > > able to set warning and critical values based on this.
> > >
> > > I'm not really sure how to implement this as it would obviously have to
> > > hold the history from another plug-in and be able to work out an average
> > > value over a timeperiod and then multiply the value until it reaches a
> > > maximum.
> > >
> > > Anyone else think that this might be useful - or has anyone already
> > > implemented this? or have any thoughts about how best to go about
> > > implementing this?
> > >
> > > Kind regards,
> > > Steve
> > >
> > > --
> > > Steve Freegard
> > > Systems Manager
> > > Littlehampton Book Services Ltd.
> > >
> > >
> > > **********************************************************************
> > > This email and any files transmitted with it are confidential and
> > > intended solely for the use of the individual or entity to whom they
> > > are addressed. If you have received this email in error please notify
> > > the system manager.
> > >
> > > This footnote also confirms that this email message has been swept by
> > > MIMEsweeper for the presence of computer viruses.
> > >
> > > www.lbsltd.co.uk
> > > **********************************************************************
> > >
> > >
> > >
> > > -------------------------------------------------------
> > > This sf.net email is sponsored by:ThinkGeek
> > > Welcome to geek heaven.
> > > http://thinkgeek.com/sf
> > > _______________________________________________
> > > Nagios-users mailing list
> > > Nagios-users at lists.sourceforge.net
> > > https://lists.sourceforge.net/lists/listinfo/nagios-users
> > >
> > > ::: Please include Nagios version, plugin version (-v) and OS when
> > > ::: reporting any issue. Messages without supporting info will risk being
> > > ::: sent to /dev/null
> >
> > -------------------------------------------------------
> > This sf.net email is sponsored by:ThinkGeek
> > Welcome to geek heaven.
> > http://thinkgeek.com/sf
> > _______________________________________________
> > Nagios-users mailing list
> > Nagios-users at lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/nagios-users
> >
> > ::: Please include Nagios version, plugin version (-v) and OS when
> > ::: reporting any issue. Messages without supporting info will risk being
> > ::: sent to /dev/null
> 




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Nagios-users mailing list
Nagios-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. 
::: Messages without supporting info will risk being sent to /dev/null