[mrtg] Re: mrtg and billing WAS: Total bandwidth usage and

Alex van den Bogaerdt alex at slot.hollandcasino.nl
Thu Oct 28 19:53:47 MEST 1999

> Chris> Someone I know of had a similar idea, except he was going to
> Chris> subtract ifinerrors and ifouterrors from each target, and then
> Chris> do a 95th percentile.
> good point.
> Chris> What about the idea of using MRTG with RRDTool, doing a RRD
> Chris> dump of the file at the end of a month, and then going through
> Chris> the data?
> when does rrdtool compress the data?  I thought that every day it went 
> through and dumped one day's worth down to the average samples, so
> that it could keep the database down to a reasonable size for forever.

No, it works different.  It recalculates the samples to normalized points
in time.  Each sample is then fed to the different archives in the database.
In other words: data from the "daily" archive is not used to calculate the
"weekly" archive and so on.
Note:  *you* define the size of each archive.  If you want to keep 5-minute
samples for two months, thats okay.

I know nothing about this "95th percentile" (guess it is not used here?)
but if you want to calculate the traffic that passed through the interface,
you may multiply the time passed by the average rate:

Average rate is in bytes per second, time is in seconds

   b   s   b
   - * - = -  -->  you get bytes if you input (bytes/sec) and (time in sec).
   s   1   1

You can do all sorts of calculations on the output of the database using
the CDEF language so maybe this will help for the 95th %ile.
It is at least able to subtract ifInErrors from ifInOctets etcetera.

 / alex at slot.hollandcasino.nl                  alex at ergens.op.het.net \
| work                                                         private |
| My employer is capable of speaking therefore I speak only for myself |

* To unsubscribe from the mrtg mailing list, send a message with the
  subject: unsubscribe to mrtg-request at list.ee.ethz.ch
* The mailing list archive is at http://www.ee.ethz.ch/~slist/mrtg

More information about the mrtg mailing list