[email protected]
[Top] [All Lists]

Question regarding draft-ietf-rtgwg-ipfrr-spec-base-05

Subject: Question regarding draft-ietf-rtgwg-ipfrr-spec-base-05
From: "Schnitter, Stefan"
Date: Fri, 23 Feb 2007 15:41:38 +0100
Hi all,

[I assume  draft-ietf-rtgwg-ipfrr-spec-base-05 - though recently expired
- is still state of discussion regarding lfa - otherwise you can ignore
the following... ]

I was looking into ipfrr opportunities for Deutsche Telekom's internet
backbone and implemented the algorithm to calculate LFA's that is given
in section 3.6 of draft-ietf-rtgwg-ipfrr-spec-base-05 (mainly to
research the different types of coverage)
I was wondering why the distance towards the destination is not stated
more clearly as a criterion to choose the LFA if there are multiple
(e.g. node protecting) candidates available. In Step 15 (page 16) it
says

 "15.  Based upon the alternate types, the alternate distances, IP
       addresses, or other tie-breakers, decide if H_h is preferred to
       P_i.alt_next_hops.  If so, goto Step 18."

Having the shortest path towards the destination - also in an ipfrr
situation - can be beneficial (within the simulation I found numerous
case where long detours occur when the distance would not be considered)
but also the distance is a tie-breaker that is easy to evaluate.
Otherwise simulations of ipfrr might more difficult (in case of DT's
backbone the distance would almost always be sufficient to choose a
unique LFA)

So - wouldn't it be good to add an "If-condition" between 14. and 15.
that explicitly chooses the candidate with minimum distance (and not
leave this to the vendors decision)?

TIA,
 Stefan

--
Dr. Stefan Schnitter
T-Systems Enterprise Services GmbH, Systems Integration, Telco Projects
& Design
Line Of Business Networks and Processes, Team Traffic Management and
Network Optimization
Phone: +49 (0) 6151 937-8521

_______________________________________________
rtgwg mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/rtgwg

<Prev in Thread] Current Thread [Next in Thread>