[69159] in North American Network Operators' Group
RE: Overflow circuit
daemon@ATHENA.MIT.EDU (Joe Chin)
Sun Mar 28 18:03:10 2004
From: "Joe Chin" <jcc-list@thenetexpert.net>
To: <nanog@merit.edu>
Date: Sun, 28 Mar 2004 15:02:14 -0800
In-Reply-To: <014901c414b2$6fec7c40$6401a8c0@stephen>
Errors-To: owner-nanog-outgoing@merit.edu
QoS mechanisms (i.e. CBWFQ/LLQ) take care of prioritizing voice over data.
So I am not worried about data. Dynamic routing (EIGRP in this case) does a
beautiful job of failing over IP traffic from one T1 to another when
required.
The objective of this exercise is to see how we can go about overflowing
voice traffic from the terrestial T1 to the satellite T1. So, this is not
really about load-balancing or load-sharing.
At this time, it looks like using H.323 gatekeeper as suggested by Patrick
Murphy maybe the most likely way to go.
> -----Original Message-----
> From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On
> Behalf Of Stephen Sprunk
> Sent: Sunday, March 28, 2004 2:35 AM
> With G.729a and cRTP, you can cram over a hundred calls into
<...>
> a single T1; what is the average and peak usage predicted for
> your deployment?
>
> While I know it's not ideal, you may want to consider using
> CallMangler's bandwidth control features to ensure voice
> traffic never exceeds the T1's capacity. That way you can
> route all voice to the T1, all data to the satellite, and not
> worry about fancy load-sharing tricks.
