Chanjing,

By using sampling, only part of the real traffic flow are exported.
In your case, one out of ten are exported. Simply by timing 10 to
your result is not a prefered scaling method, because the sampling
is based on flow not bytes, by #flow*10 you may get the close
enough flow number, but #bytes*10 is way far from the real traffic
bytes. The prefered way is to scale based on the real interface
out/in byte counter, as MRTG/RTG etc provides.

Regards,
Eric

[EMAIL PROTECTED] wrote:


Today's Topics:

   1. the netflow sampling problem (Chanjing Han)
   2. flow-fanpout problem (Li pengfei)
   3. flow-fanout question (wangyan(cernoc))
----------------------------------------------------------------------

Message: 1
Date: Mon, 11 Apr 2005 09:07:39 +0800
From: "Chanjing Han" <[EMAIL PROTECTED]>
Subject: [Flow-tools] the netflow sampling problem
To: "flow-tools" <[EMAIL PROTECTED]>
Message-ID:
        <[EMAIL PROTECTED]>
Content-Type: text/plain;       charset="gb2312"

hi:
  Have you meet the netflow sampling problem?  I get the netflow data by 10:1 
sampling, In my project ,I deal with the traffic by 10*traffic when puting the 
traffic in my RRD . But I find that the traffic has a big discrepancy with the 
traffic by MRTG.



����������������Chanjing Han
[EMAIL PROTECTED]
��������������������2005-04-11


_______________________________________________ Flow-tools mailing list [EMAIL PROTECTED] http://mailman.splintered.net/mailman/listinfo/flow-tools

Reply via email to