The MPLS-OPS Archive

[sthaug@nethelp.no]

> When you talk about pure Layer 2 networks and their scalability, are
> you also referring to layer 2 circuits built around AToM and/or
> L2TPv3?

In the current thread I was mostly talking about networks made from
Ethernet switches. Here you get into scaling problems because of the
size of address tables, size of VLAN space, spanning tree issues, and
a host of other issues (e.g.: If *one* end of a VLAN connection from
A to B goes down, how do you communicate this to the other end in a
standardized manner? There is no equivalent of AIS, RDI etc.)

For AToM/L2TPv3 (and excluding VPLS) you basically have pure point
to point connections which are "pipes" - put a packet in at one end
and it comes out at the other end. No MAC address learning needed,
no explicit connection to the size of the VLAN space, redundancy is
normally handled by L3 routing protocols instead of STP. From this
point of view, AToM/L2TPv3 is more scalable than pure Ethernet L2.
On the other hand, AToM and L2TPv3 are *more complex* technologies
than basic Ethernet L2, which of course results in its own set of
problems.

> Can you recommend a few starters to read up on the scalability
> issues related to layer 3 networks and layer 2 circuits (built
> around AToM)?

Not sure if such literature exists yet. You may be able to find
something in presentations from various vendors (try the Cisco 
Networkers presentations, for instance).

Steinar Haug, Nethelp consulting, sthaug@nethelp.no

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml