If you don't want to keep 30 days worth of individual samples, determine how many samples make up a month (5 minute samples, 30 days = 8640), take 5 percent of that (432), save off that many (432) samples, each day recalculate the top (432) to be saved. At the end of the month, you have the top 5% samples. The lowest one = 95th percentile. We ran into this not wanting to redo 100,000+ RRDs to keep 30days of data.
--------------------------------------------------- Kathy Kessey [EMAIL PROTECTED] [EMAIL PROTECTED] 650.556.5087 On Thu, 14 Feb 2002, David Gabler wrote: > > I have done this. > > It is really called 95th Percentile > > The psudocode is to > rrdtool fetch all the data for the past 30 days. > figure out how many data points you have > pop off the top 5% > pop off the top# and there you go. that is your 95th Percentile. > > David > > > -----Original Message----- > > From: Geert van der Ploeg [mailto:[EMAIL PROTECTED] > > Sent: Thursday, February 14, 2002 8:46 AM > > To: [email protected] > > Subject: [rrd-users] Calculating a 5%-rule > > > > > > > > Hi all, > > > > I'm implementing a sort of bandwidth-usage billing system, > > using RRDtool. > > > > <short mode> > > How to calculate what value is the highest, after ignoring > > the highest 5% ? > > </short mode> > > > > -- > Unsubscribe mailto:[EMAIL PROTECTED] > Help mailto:[EMAIL PROTECTED] > Archive http://www.ee.ethz.ch/~slist/rrd-users > WebAdmin http://www.ee.ethz.ch/~slist/lsg2.cgi > > -- Unsubscribe mailto:[EMAIL PROTECTED] Help mailto:[EMAIL PROTECTED] Archive http://www.ee.ethz.ch/~slist/rrd-users WebAdmin http://www.ee.ethz.ch/~slist/lsg2.cgi
