So if T-MPLS is a look-out for trouble, for much-bigger-than-metropolitan network architectures, I am down to 3 choices. Let's assume PBB-TE is not yet widely implemented and let's assume that there are few automated provisioning interfaces designed for MEF9+14 equipment, then it doesn't leave much other choice but VPLS does it not ? F. On 2010-11-13, at 5:38 PM, Mark Smith wrote:
On Fri, 12 Nov 2010 08:30:19 -0500 Francois Menard <francois@menards.ca> wrote:
I'm embarking on a new project which involves a large scale MAN network where ultimately, the objective is to carry QinQ, while at the same time delivering services over IPv6.
The objective is to support jumbo frames on all interfaces, at least to carry QinQ standard-size ethernet frames, but ideally as large as possible
There seem to be 4 approaches to do this.
a) The IEEE PBB-TE approach - but little implementations. b) The MEF9+14 approach, mature, but manual provisioning c) The VPLS approach, concerns with too much manual provisioning. d) The T-MPLS approach, concerns with maturity
The objective is to support the functionality not only in the CORE, but also on cost effective multi-tenant & redundant customer CPEs.
I have not seen a, or b or d supported in a low-cost customer CPE.
I am currently favouring c, for reasons of maturity and wide implementation, but may be missing on recent progresses in the b) land.
Any thoughts ?
Any published IETF material on the topic ?
I'd avoid T-MPLS -
"Uncoordinated Protocol Development Considered Harmful"
http://tools.ietf.org/search/rfc5704
Regards, Mark.