[35225] in North American Network Operators' Group
Re: Does peering at an ATM MAE require a router that does ABR?
daemon@ATHENA.MIT.EDU (Steve Feldman)
Thu Mar 1 17:06:19 2001
Date: Thu, 1 Mar 2001 13:59:46 -0800
From: Steve Feldman <feldman@twincreeks.net>
To: nanog@merit.edu
Cc: tek weeny <tekweeny@hotmail.com>
Message-ID: <20010301135946.B97064@twincreeks.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <200103012136.f21La3L32712@ptavv.es.net>; from oberman@es.net on Thu, Mar 01, 2001 at 01:36:03PM -0800
Errors-To: owner-nanog-outgoing@merit.edu
On Thu, Mar 01, 2001 at 01:36:03PM -0800, Kevin Oberman wrote:
>
> In the past, the MAE ATM switches ran ABR where you specified the MCR
> and traffic was policed at a PCR of 2xMCR, although almost no one at
> the MAEs actually used ABR on their connections. Almost everyone used
> VBR-nrt and many did not even realize that the use of ABR could have
> been beneficial.
To be even more precise, MAE-ATM runs ABR
nternally, but didn't speak the protocol
over the UNI to the routers.
In this mode, the switches being used would
emulate VBR-nrt but actually do some input
buffering and shaping to smooth out bursts.
This came in very handy for older router interfaces
which couldn't shape on their own.
I never had an ABR-speaking router to try it with.
> With "best effort" PVCs, this really becomes moot although these are
> still technically ABR.
Agreed, though I'm not famiiliar with details of the
"best effort" service. And I'm pretty sure the reserved
and policed service is still there for folks who still
want to use it.
Steve
(former target-carrier)
