[rrd-users] Re: New operator to CDEF proposal

[Glenn MacGregor]

That would be true if we were gareented that the time the samples are taken
are all equal distances.  We still need to account for unknown samples.

Here is a better explanation (example)
Time        Value
0                1
5                1
9                0
20              1

We need to make some assumptions here, between 5 and 9 somewhere the value
changed to 0 could have been right after we polled at 5 or right before we
poll at 9.  So I assume that it is at 7 ((9-5)/2) + 5 (middle).  Same with
the 9 to 20 case so I assume it is 14.5.

So the value is 1 from 0 to 7 and 14.5 to 20, total time it is 1 = 12.5 of
the 20.  Total time it was 0 is 7.5. The calulation would be
(20 - 7.5)/20 = .625  ==> Total up percentage is 62.5%

----- Original Message -----
From: "Gonzalo Arana" <garana at uolsinectis.com.ar>
To: "Glenn MacGregor" <gtm at highstreetnetworks.com>
Sent: Wednesday, January 22, 2003 4:29 PM
Subject: Re: [rrd-users] New operator to CDEF proposal


>
> mmm...
>
> that's equal to (apart from diferences due to finite numerical
> precision):
>
> #TrueSamples/#TotalSamples
>
> #TrueSamples: can be obtained via CDEF: (sum over all values).
> #TotalSamples: yo know it since you built rrd.
>
> Note: you may only graph rra, NOT rrd.
>
> Hope it helps,
>
> On Wed, 2003-01-22 at 14:00, Glenn MacGregor wrote:
> > Hi All,
> >
> > I have not been using rrdtool for too long, so this maybe unnecessary,
if so tell me know how I can do it with the existing operators.
> >
> >
> > I am trying to calculate availability for a device based on a true or
false value.  The availability is a percentage 0 - 100%.  The only way I can
think of to calculate it would be:
> >
> > ((Total # samples * (endtime - starttime)) - (# of false samples *
(endtime - starttime)) / (Total # samples * (endtime - starttime))) * 100
> >
> > Example:
> > Ten samples every 5 minutes
> > Time        Value        Availability        Function
> > ---------    -----------        --------------        --------------
> > 0                1                100%
((1*(0-0))-(0*(0-0))/(1*(0-0))*100               (Could be a problem here
Div by 0)
> > 5                1                100%          ((2*(5-0))-0/2*(5-0)*100
= 100
> > 10              1                100%
> > 15              1                100%
> > 20              0                80%            (5*20)-(1*20)/(5*20) =
80/100 = .8 * 100 = 80%
> > 25              1                83.3%
> > 30              0                71.4%
> > 35              0                62.5%
> > 40              1                66.7%
> > 45              1                70.0%
> > 50              0                63.7%
> >
> >
> >
> > Is there a way to do this in the current CDEF?  If not I can add it to
the list of functions if you think this is the correct way to do something
like this?
> >
> > Or is there another way to make this happen within rrd?
> >
> >
> > Thanks in advance
> >
> >         Glenn MacGregor
> > --
> > 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
> >
> >
> --
> Gonzalo Arana
> Ingenieria de Portal
> UOLSinectis
>
> Florida 537 Piso 6, Buenos Aires, Argentina
> +54-11-4321-9110 ext 2543
> http://www.uolsinectis.com.ar/
>
>
>

--
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