[190199] in North American Network Operators' Group
Re: 1GE L3 aggregation
daemon@ATHENA.MIT.EDU (Mark Tinka)
Sat Jun 18 07:07:08 2016
X-Original-To: nanog@nanog.org
To: Baldur Norddahl <baldur.norddahl@gmail.com>, nanog@nanog.org
From: Mark Tinka <mark.tinka@seacom.mu>
Date: Sat, 18 Jun 2016 13:04:49 +0200
In-Reply-To: <CAPkb-7DXiAkEoW_fW55Sbw_0hq1HUFoVPz-PdLwDG+Ouc-Nvfg@mail.gmail.com>
Errors-To: nanog-bounces@nanog.org
On 16/Jun/16 21:36, Baldur Norddahl wrote:
> Hi
>
> If I need to speak BGP with a customer that only has 1G I will simply m=
ake
> a MPLS L2VPN to one of my edge routers. We use the ZTE 5952E switch wit=
h
> 48x 1G plus 4x 10G for the L2VPN end point. If that is not enough the Z=
TE
> 8900 platform will provide a ton of ports that can do MPLS.
>
> The tunnel is automatically redundant and will promote link down events=
, so
> there is not really any downside to doing it this way on low bandwidth
> peers.
Personally (and at work), I stay away from such topologies. Centralizing
IP connectivity like this may seem sexy and cheap in the start, but it
has serious scaling and operational issues down the line, IMHO.
We push IP/MPLS all the way into the Metro-E Access using a team of
Cisco ASR920's and ME3600X's. The value of being able to instantiate an
IP service or BGP session directly in the Metro-E Access simplifies
network operations a great deal for us. Needless to say, not having to
deal with eBGP Multi-Hop drama does not hurt.
Centralizing is just horrible, but that's just me. The goal is to make
all these unreliable boxes work together to offer a reliable service to
your customers, so making them too inter-dependent on each other has the
potential to that away in the long run.
Mark.
