[ILUG] [Possibly OT] Network Traffic Analysis and Interface Control

[Hamilton, David (TSG C&I)]

The link is being used for a single server to server encrypted transfer,
so WRED doesn't make sense unfortunately.
The link is also not available currently as a per-MB billed link, so
that's not an option either.

I was thinking of something that could monitor the output of tethereal
and act upon certain criteria...
... Still googling...

	D.

-----Original Message-----
From: Paul Jakma [mailto:paul at clubi.ie] 
Sent: 22 February 2006 17:10
To: Hamilton, David (TSG C&I)
Cc: ilug at linux.ie
Subject: RE: [ILUG] [Possibly OT] Network Traffic Analysis and Interface
Control

On Wed, 22 Feb 2006, Hamilton, David (TSG C&I) wrote:

> No. It's more a case of having to use an unreliable, high-cost 
> transmission medium which gives a highly variable bandwidth.
> When the effective bandwidth does drop below useable levels, we have 
> usually found tons of TCP re-transmissions, and would therefore like 
> to use these as a criterion for shutting down the line due to lack of 
> financial viability of the session.

Ah, bit more sensible :).

If this link is carrying multiple TCP sessions, try configuring WRED on
the output interface, should make a difference. Some recent TCP
implementions have better congestion/contention algorithms (e.g BIC)
which could also help, if that's an option.

Financial viability: Presumably you've already negotiated per-MB billing
on this link? (ie not billed by 'session-connected' time).

Your original question: No idea, sorry.

regards,
-- 
Paul Jakma	paul at clubi.ie	paul at jakma.org	Key ID: 64A2FF6A
Fortune:
Hildebrant's Principle:
 	If you don't know where you are going, any road will get you
there.