Dear all,
with
this mail we would like to
focus the attention on the disjoint path computation
issue, within the context of inter-area/AS TE.
We believe that not addressing
the disjoint path computation issue from the start (when looking at inter-area/AS TE) would
be quite problematic,
since an approach that works
for a single inter-area/AS path
may not be
easily applicable/extendible for diverse path
computation. This observation has been made earlier
on ML discussions after the Seoul
meeting, but there was no definitive conclusion at that time (http://ops.ietf.org/lists/ccamp/ccamp.2004/msg00336.html
and follows). In fact,
in our view, the right way to approach
this issue would be to
develop an approach that that
does not provide a solution to setup a single path and then attempts
to extend that solution for
the diverse path setup case.Rather, a solution
must provide a mechanism to setup disjoint
paths, with the single path setup being
a
particular case. This is because disjoint path setup is
quite likely one of the
more important aspects of inter-region TE, that is important for
a no. of applications,
as pointed out in the requirements
drafts.
We defined such an
approach in our draft “draft-dachille-inter-area-path-protection-
00.txt”, discussed in Seoul (59th IETF)
and it received many positive comments (see the
minute of the
meeting): in many comments it has been
marked as an important work, since it addressed
a hot issue within the Inter-area/AS
topic. Following, the draft has
been discussed on the CCAMP
ML for about two months Recently
we have reviewed
our draft (namely draft-dachille-diverse-inter- region-path-setup-00), incorporating
the feedbacks received at Seoul and via the CCAMP discussions
that followed. Since there are significative changes, we recently asked
for a slot during the 60th
IETF, just to discuss the
new version (http://ops.ietf.org/lists/ccamp/ccamp.2004/msg00720.html).
We would like to
address two main themes:
i) clarifing why an approach
to achieve diverse path setup from
the
start is now not
considered in the inter-region
TE solutions,
notwithstanding it seems to be the right
approach to the issue, as
observed in prior ML discussion
and in requirement drafts.
ii) Since we believe that draft-dachille
satisfies all the
requirements for work to be discussed at a WG meeting
(http://ops.ietf.org/lists/ccamp/ccamp.2004/msg00746.html),
we would appreciate clarification about the reasons that led
to the
exclusion of this work from
the SD agenda.
Thank you for
your kind attention; we hope
that the discussion about
this issue could continue
in a productive manner,
just as usual.
Best Regards,
Alessio D’Achille