[89016] in North American Network Operators' Group
Re: Identify amount of traffic to special IP address in Radius
daemon@ATHENA.MIT.EDU (Christian Kuhtz)
Wed Mar 1 00:28:47 2006
In-Reply-To: <20060301045957.45927.qmail@web53615.mail.yahoo.com>
Cc: NANGO <nanog@merit.edu>
From: Christian Kuhtz <kuhtzch@corp.earthlink.net>
Date: Wed, 1 Mar 2006 00:28:14 -0500
To: Joe Shen <joe_hznm@yahoo.com.sg>
Errors-To: owner-nanog@merit.edu
On Feb 28, 2006, at 11:59 PM, Joe Shen wrote:
>> why in the world
>> would you want to do something like that rather than
>> have another
>> device generate flow records which you then can
>> correlate with RADIUS
>> accounting data?
>>
>
> The reason is the cost of system building. As there
> are a lot of broadband subscribers, if we want to
> corelate subscriber with web site they visit we have
> to make Radius allocate a fixed IP to a special
> subscriber. Netflow based accounting is costy and we
> could not guarantee its accuracy.
I would submit to you that getting such functionality built and
operating via RADIUS is expensive as well. If you have a lot of
broadband subscribers, your BRAS will already be busy doing what it
does 'best': terminating subscriber sessions. Plus, you have to get
your vendor to implement such a feature.. I haven't seen something
like what you describe in the years of looking at BRAS requirements.
But, that doesn't mean it doesn't exist in a scalable fashion.
> In fact, we only need to identify a small set of IPs
> which should not be considered in accounting.
If you need to generate IP accounting for just a small set of IPs,
the above engineering challenge you describe is very straightforward
for capturing flow records.
YMMV, of course. ;-)