[j-nsp] MPLS transport questions on l2circuit vs proto connections

[Alexander Koch]

Everyone,

I have some questions about the two variants here and I am
looking for the abstract way of doing things for a L2 transport
service on a Juniper network. I am sure several here on this
list can share some insight with ease.

With protocol connections you have the most fine- grained
traffic engineering, where with l2circuit you have very little
basis for manipulation per actual transport - can I say that
and not blush for being terribly wrong?

How do I actually do traffic engineering in any good way on
l2circuits? I can build ldp- tunnels, but that is it?

Further to that can I apply policers on 'family ccc' (sub-)
interfaces, and will they actually work? I know I can configure
them, but will they just do what I expect of them (like on L3)?

Also can I limit the amount of Mbit/s any remote-int-switch
or a l2circuit gets per individual transport? I see 'bandwidth'
in the l2circuit config as an option, but I guess I need a ton
of additional config or traffic classes, any pointers or key
words to look for on this one?

In general, why would I want to use l2circuits vs old- style
proprietary connections? LDP does interop with Cisco and Foundry
etc, but I have less granularity when there is re- routing needed.
While with proto connections I have all the power in the world
but a ton of additional config (LSPs). Which one would you recommend?

Disclaimer:
Any errors here are my own, please correct them. I have used both
variants before, but never had to compare them, and surely do not
know all the (newer) options / features.

Thanks lots,
Alexander