[69139] in North American Network Operators' Group
Re: Overflow circuit
daemon@ATHENA.MIT.EDU (Patrick Murphy)
Sat Mar 27 11:13:49 2004
From: "Patrick Murphy" <pjm@nfld.net>
To: "Alexei Roudnev" <alex@relcom.net>,
"Mailing List Subscriptions" <jcc-list@thenetexpert.net>,
<nanog@merit.edu>
Date: Sat, 27 Mar 2004 12:42:59 -0330
Errors-To: owner-nanog-outgoing@merit.edu
It works, I am doing MPLS VPN's using 3662's and running VoIP toll bypass.
Customer is very pleased with the network.
Patrick
----- Original Message -----
From: "Alexei Roudnev" <alex@relcom.net>
To: "Patrick Murphy" <pjm@nfld.net>; "Mailing List Subscriptions"
<jcc-list@thenetexpert.net>; <nanog@merit.edu>
Sent: Saturday, March 27, 2004 4:09 AM
Subject: Re: Overflow circuit
> VoIP over satellite? I am very sceptical about it. Better, forget such
idea.
>
>
>
> >
> > You may want to look at using H.323 gatekeepers with CAC (Call Admission
> > Control).
> >
> > Here is a link to a Whitepaper on this Subject.
> >
> >
>
http://www.cisco.com/en/US/partner/tech/tk652/tk701/technologies_white_paper09186a00800da467.shtml
> >
> > Patrick
> >
> > ----- Original Message -----
> > From: "Mailing List Subscriptions" <jcc-list@thenetexpert.net>
> > To: <nanog@merit.edu>
> > Sent: Friday, March 26, 2004 7:54 PM
> > Subject: Overflow circuit
> >
> >
> > >
> > >
> > > I am looking for advice on technique or products that can solve the
> > > following challenge ...
> > >
> > > Two private line T1's between A and B - one terrestial T1 with >200 ms
> > RTT,
> > > the other T1 is over satellite with ~500 ms RTT. The circuits are
being
> > used
> > > for mixed VoIP (70%) and data (30%) applications. To achieve optimal
> voice
> > > quality, we want to route all VoIP calls over the terrestial T1 until
it
> > is
> > > "full", then divert all subsequent VoIP calls over the satellite T1
(**
> > > while existing VoIP calls continue to be routed over the terrestial
T1).
> > >
> > > So it looks like I need per-flow (based on protocol, src IP, dst IP,
src
> > > port, dst port) routing. It looks like MPLS Traffic Engineering can do
> the
> > > job. Is there anything else that can it with less complexity?
> > >
> > > Ideas or recommendations?
> > >
> > >
> > > Regards,
> > > Joe
> > >
> > >
> >
>
