[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Diverse path failure and optimality in draft-dachille-inter-area-path-protection



hi,

Vishal Sharma wrote:
Hi Dimitri,

Thanks for your thoughts.

However, as Fabio rightly observed, it is very critical that we
distinguish between characteristics that are intrinsic (and common)
to _all_ distributed path computation approaches from those that are
specific to the ARO approach.

i don't think i'm saying that the ARO approach should solve all the problems i'm just trying to see the problems that the ARO approach solves and to which extend/conditions


The points you have mentioned below are inherent in _any_
distributed path computation paradigm (and indeed in any
distributed signaling/resource reservation paradigm), as I explain below,
and are therefore not drawbacks that arise due to particular
mechanisms/methods proposed in the ARO approach.

once again, this is not the way these points were raised (they are not saying there are no problems with existing methods) but to which extend the proposed ARO approach enforce them (or avoid them)


as last philosophical word i'd like to point the following:
<ftp://ftp.rfc-editor.org/in-notes/rfc3439.txt>

it details several principles (but sure we've already cross some of the boundaries this document mentions) that can be useful in re-positioning the whole "optimization" debate

thanks,
- dimitri.
Comments in-line.

-Vishal


-----Original Message-----
From: owner-ccamp@ops.ietf.org [mailto:owner-ccamp@ops.ietf.org]On
Behalf Of Dimitri.Papadimitriou@alcatel.be
Sent: Monday, April 26, 2004 6:47 AM
To: ugo monaco
Cc: ccamp; Zafar Ali
Subject: Re: Diverse path failure and optimality in
draft-dachille-inter-area-path-protection



hi, by discussing the proposed method there seems to be three issues
that make the method questionable in terms of guarantee to deliver what
it intends to deliver, its usability (the time validity of the path is
not guaranteed) and its applicability in terms of objective initially
targeted wrt to the network topology

1) imagine three areas decoupled computation as explained at their
respective ingress, with ARO method; how the third computation element
(tail-end) is aware of srlg's that may affect a link selected in the
head-end area

example: link 1 is selected in area 1 (head-end) with srlg 1, link 2 is
selected in in area 2 with srlg 2, and link 3 in area 3 (tail-end) with
srlg 3 and 1 (since the tail end doesn't know that srlg 1 is associated
to link 1 in addition to its association to link 3 even if it knows that
the link 1 has been selected for the ARO) the problem is that you can
not retrieve such kind of error (except but how practical is it if one
start cumulating all this information between computation points)


- First, if we are talking about IGP areas, the problem is a non-issue.

(They would be under the control
of a single provider, and the provider can reasonably be expected to
assign distinct, globally unique SRLGs to the links in different areas.)

- If we are talking about AS's, the problem is _universal_, and all
schemes related to restoration have to deal with it. In particular,
_any_ scheme that does distributed path computation has to deal with it.

So the question is unrelated to the ARO scheme, but a broader one
about how different providers assign SRLG's to resources.

This is actually a very important issue, but outside the scope of the
ARO document, and not one that I have seen any comprehensive approach to,
yet.

Perhaps others on the list (carriers?) can comment on it.


2) resource contention, the secondary path may never be established
since the computation point as *no* capability to make any reservation
on it (except from the first segment) since by definition "disjoint" -
it simply becomes a kind of "best effort" secondary path (in the sense
use it if no other reservation are making use of these links)


As Fabio has very nicely explained in
his email,
http://ops.ietf.org/lists/ccamp/ccamp.2004/msg00553.html

resource contention is a fact of life (read unavoidable) in any distributed
scheme (unless you assume _instantaneous communication_ between the
distributed entities), and is prevalent in every distributed
protocol/mechanism from RSVP-TE to mechanisms for diverse path
computation and setup, such as PCE-based, XRO-based, and, of course,
ARO-based
schemes.

Not sure why you thought this is specific to the ARO proposal.


3) the method seems to raise additional issues when the number of
potential entry point for the secondary disjoint path increases, at each
step of the computation (otherwise the method wouldn't provide what it
intends to deliver)


First, not sure what "additional issues" you are referring to. Can you
provide specifics?

In any case, every path computation scheme (distributed or otherwise) has to
handle
the case where there are multiple entry points for the secondary (and
primary) paths at a given area/AS. Again, nothing surprising here or
specific
to the ARO scheme in particular.

When there are many entry points, any scheme has to have a way to select
only one (or some number of them), and any one of several criteria may be
used to make that selection.

Of course, one can consider then the relative merits of the different
ways of making that selection.



ugo monaco wrote:



Dear ccamper, Zafar, Dimitri,

as anticipated in the recent email by Vishal we are addressing several
important comments collected at Seoul regarding the joint-selection of
diverse paths with ARO, i.e. the approach proposed in


<http://www.ietf.org/internet-drafts/draft-dachille-inter-area-pat h-protection-00.txt>.


As Vishal did I summarized your comments to ensure that we rightly understood inputs, and to help people on the ML follow and contribute to the discussion. Comments from others who have feedback are welcome, and much appreciated.

Please let me know if you had any additional comments
as well. We will take these into account in providing our
responses, and, later, in updating the document.



Thanks again for your feedback on our draft during the Seoul meeting.
Best Regards.

Ugo Monaco




++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++++++++++++++++++++++++



Zafar's questions:

i) What happens when the setup of the diverse path fails, or there is a
failure on it after it has been set up?

ii) Relationship of this scheme to PCS/PCE approach of JP?


Dimitri Papadimitriou


i) Your question was about why we are trying for optimality

with a joint


path selection scheme, when it is not possible, especially as

the number


of AS's or areas along a path increase, to achieve the *global* optimum.
Your other comment was that we should mention this somewhere in the
document.

Please note that in response to this last question Fabio

Ricciato listed


many
advantages of a joint computation (with ARO) of inter-area/AS

paths in a


prevoius email posted on the ML "About optimality of inter-AS parallel
computation in draft-dachille-inter-area-path-protection".
We hope that Fabio rightly addressed your input and we will appreciate
further comments and notes.





-- Papadimitriou Dimitri E-mail : dimitri.papadimitriou@alcatel.be E-mail : dpapadimitriou@psg.com Webpage: http://psg.com/~dpapadimitriou/ Address: Fr. Wellesplein 1, B-2018 Antwerpen, Belgium Phone : +32 3 240-8491







-- Papadimitriou Dimitri E-mail : dimitri.papadimitriou@alcatel.be E-mail : dpapadimitriou@psg.com Webpage: http://psg.com/~dpapadimitriou/ Address: Fr. Wellesplein 1, B-2018 Antwerpen, Belgium Phone : +32 3 240-8491