[rrd-users] Re: Calculating average when aggregating

Geert van der Ploeg rrd at geert.triple-it.nl
Mon Aug 5 15:11:30 MEST 2002

> > I would say that's exactly what you want, isn't it?
> > You want the average-calculating to start when the first graph gets its
> > first non-unknown value.
> >
> > So, If the first non-unknown value will only start after 1/4 of the time
> > the graph, the average will be calculated over the remaining 3/4 of the
> > time.
> >
> > If you do not want that (but you want to start the average-calculating
> > the start of the graph), you could just take the same CDEF as the
> > nice-line-graph, thus bringing you back in the same position as where
> > began.
> >
> O.K., let's try to make it clear:
> 1. 6 months ago I started the first rrd.
> 2. Today the second rrd starts.
> 3. To get a nice graph, I eliminate the "*UNKNOWN*"s. This will lead to
> an average calculation for over the last 12 months (if the graphs goes
> over 12 months), so I do a second CDEF.
> 4. This leads me to the situation (I guess it's like that, if not I'd be
> glad) that the average calculation starts today. I would like it to
> start 6 months ago, when the first rrd started. My guess comes from the
> behaviour of the aggregated graph, if things are different, I'd be fine.

What is your way of averaging?
We should be able to determine what the outcome should be, following the

If the average does not start at the beginning and doesn't do the
calculation right, a (nasty) solution would be to change the UNKNOWN value
to the value of the 'known' DS.
Thus, you get an average which is equal to the only known DS, as long as the
other, UNKNOWN DS is unknown.
That would be something like: ds1,UN,ds0,ds1,IF (where ds1 is the unkwown,
newest DS, and ds0 the DS that started 6 months ago)



Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi

More information about the rrd-users mailing list