[rrd-users] Experiance in HW forcast

[j.halifax2 at seznam.cz]

Hi all,




In my experience, for forecasting it is necessary to include some 

"intelligence" about the nature of data. The "blind" forecasting didn't

provide me satisfactory results. For example, in my pro-active system 

monitoring Multi-ISP operation (in 2001) alerting and predicting problems

I had to include calendar, working days and public/school holidays etc., 

and then the system was able to disclose problems coming successfully.




I see the forecasting as quite strongly field-dependent matter.




Leos





---------- Původní zpráva ----------
Od: Tobias Oetiker <tobi at oetiker.ch>
Datum: 13. 5. 2012
Předmět: Re: [rrd-users] Experiance in HW forcast
"Hi Michael,

Today Michael Schwartzkopff wrote:

> Hi,
>
> In the last week I read into the HW forecasting and the implementation in 
the
> rrdtools.
>
> After reading a lot there are still some questions open:
>
> 1) How do I choose the initial values for the fitting parameters alpha, 
beta
> and gamma?
>
> in my experiments alpha behaved nicely and it was no problem to find the
> optimal alpha. But the least square function of gamma in the range of ]0,1
[
> seems to be fractal. So it will be extremely difficult to optimize the
> parameters.
> Anybody else with this experience?
>
> In all samples in the net nobody explains why he choose exactly this
> parameters. Why?

take some high resolution sample data worth several seasons (weeks)
and then try out the different parameters ... note some are very
sensitive ... the best way to get your config right is to tune with
real data ...

> 2) Is there any possibility to optimize the fitting parameters to recent 
data?
> My idea would be to optimize the fitting parameters, create the RRDs and 
to
> optimize the RRDs from time to time. Any ideas or comments?

you can rrdtool tune the parameters ...


> Are there any experts on the list? I would like to start a discussion 
about
> making the use of the forecasting in RRDs as simple as creating a plain 
graph.

there is very little discussion on this toppic, but maybe this will
start it ...

cheers
tobi


> Greetings,
>
>

-- 
Tobi Oetiker, OETIKER+PARTNER AG, Aarweg 15 CH-4600 Olten, Switzerland
http://it.oetiker.ch(http://it.oetiker.ch) tobi at oetiker.ch ++41 62 775 9902 
/ sb: -9900

_______________________________________________
rrd-users mailing list
rrd-users at lists.oetiker.ch
https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users
(https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users)"
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/rrd-users/attachments/20120513/d2870b1b/attachment.htm