[mrtg] Re: SLA monitoring, 95th percentile rule

Rainer Bawidamann Rainer.Bawidamann at rz.uni-ulm.de
Tue Jan 16 19:18:10 MET 2001

In article <NMEPKNCKBFAKHIECDKCJGEFMCAAA.swen.wulf at solidsystems.com>,
	swen.wulf at solidsystems.com ("Swen Wulf") writes:
> We have a SLA in place that says that we will take snapshots of the
> bandwidth every 5 minutes over a month. (Hello MRTG !) At the end of the
> month, the data will be sorted and the top 5% will be discarded (okay...now
> what). The next highest bandwidth snapshot will be taken to calculate a
> basis to charge the customers.

You need to change the rrd file setup of mrtg: mrtg creates the rrd file
so that the measured values (better: consolidated values) are stored for
twp days. Older values are only available as 30min or 2hour or 1day
averages. You need to store 30days of 5 minute samples in the rrd files.
Change mrtg (look for RRDs::create) or create the rrd files manually.
(you will need to store around 8640 5 minute samples)

At the end of the month do a 'rrdtool fetch file AVERAGE -s -30d' (or
use -s and -e to set the start and end of the month explicitly). Sort
the returned values and store the biggest 5% of it (around 432). The
smallest of these is the 95th percentile.

> is RRD storing all the values and I just need to learn how to program
> in perl and use the rrd format to get to that data.

perl would be a good choice but maybe a bit hard for a beginner. Just
try it and post your code, we will help you ;-)


Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi

More information about the mrtg mailing list