RE: Aberrant behaviour -- any experience?

Date view Thread view Subject view Author view

Subject: RE: Aberrant behaviour -- any experience?
From: Stanislav Sinyagin (ssinyagin@yahoo.com)
Date: Tue May 21 2002 - 16:56:53 CDT

--- Jake Brutlag <jakeb@microsoft.com> wrote:
> > I already reduced it to 2 failures in 30-minute window.
> 
> You want to go the other direction; the default is 7 or more violations
> in a window of 9 observations produces a failure. If you changed this to
> something like 2 or more violations in a window of 6 observations, this
> means that failures will occur more frequently (lower bar defines a
> failure). If there is something in the documentation which suggests the
> opposite, please let me know and I'll correct it.

no, that was my own experimenting. I thought that it's not really interesting 
to know about a failure which happened half-hour ago. I knew there's a price 
for knowing it sooner, I didn't expect it to be that high...

> > As far as I understood from the manual, aberrant-reset option 
> > does not  re-apply the algorithm to the existing data, and I can see
> the effect 
> > only few days later?
> 
> The problem is that the data for the algorithm to 're-learn' is not
> guaranteed to be available. In your case, you are preserving 2 weeks of
> fine-resolution historical data. But often the fine-resolution data
> (5-min averages) is discarded after a day or so in favor of consolidated
> data, which can't be used.

Well, anyway any available date is better to use. And the one tuning 
the settings might set the fine-resolution RRA to a higher size for the time 
of tuning. 

It's late evening here, I'll look into the sources with a fresh head and 
try to decide what's more worth implementing -- a perl script or a direct 
hack into the sources.

Cheers, 
Stanislav

__________________________________________________
Do You Yahoo!?
LAUNCH - Your Yahoo! Music Experience
http://launch.yahoo.com

--
Help        mailto:majordomo@net.doit.wisc.edu and say "help" in message body
Unsubscribe mailto:majordomo@net.doit.wisc.edu and say
"unsubscribe flowscan" in message body
Archive     http://net.doit.wisc.edu/~plonka/list/flowscan/archive/

Date view Thread view Subject view Author view

This archive was generated by hypermail 2b25 : Tue May 21 2002 - 16:57:44 CDT