The MPLS WG Archive

[David Charlap <David.Charlap@marconi.com>]

Haseeb Budhani wrote:
> 
> I'd like to get help on the following three issues:
> 
> 1. Is it possible to have different types of subojects -- IPv6 and
> IPv4 -- in an ERO object ?? In case I missed it, the draft doesnt
> allow or explicitly forbid this ...

it is permitted.  Of course, when a router identifies a v6 subobject as
the next-hop, it must be able to determine the next-hop address needed
to reach that v6 address (probably via the local routing table.)

> 2. If local repair is in place, doesnt that basically counter the
> whole idea of having explicit routes ?? If CSPF or some other
> parallel mechanism is in place, when would one want to employ local
> repair ??

Local repair is only used in the case where the ERO-specified path is
temporarily broken.  It exists to temporarily find another path (which
may not be optimal) in order to keep data-plane traffic flowing while a
new ERO is generated to replace the failed one.

> 3. It would seem unlikely that a label subobject be present in the
> Route Record Object in a Path message. If there is a label subobject
> in a Path message, should this be flagged an error ??

Why?  What would break if it was present?  What purpose is served in
generating errors and tearing down state because something of no
consequense appears out of place?

-- David