[ndnSIM] Best Route acting weird

[Junxiao Shi]

Hi Sabet


Thank you so much. You're right!! I guess this shows what you said:


> 10s 0 nfd.BestRouteStrategy2:afterReceiveInterest(): [DEBUG]
> /prefix/%FE%07?ndn.InterestLifetime=2000&ndn.Nonce=3530243069 from=259
> newPitEntry-to=257
> 14s 0 nfd.BestRouteStrategy2:afterReceiveInterest(): [DEBUG]
> /prefix/%FE%07?ndn.InterestLifetime=2000&ndn.Nonce=85550101 from=259
> newPitEntry-to=257
>

These two lines clearly indicate the consumer retransmission is treated as
new Interest by best-route v4.

So, the helper does not really fail a link. Isn't failing/uping a link what
it actually supposed to do?

LinkControlHelper fails a link by setting it's loss rate to 100%. This is
correct.
What's missing is: the face state won't report as DOWN. This can be
achieved either by BFD, or by monitoring the underlying transport and
change state directly https://redmine.named-data.net/issues/3352 .
The equivalence of the latter approach in ndnSIM is to set the transport
state as DOWN in LinkControlHelper. Let's see how ndnSIM developers think
about this.

What is RetxSuppressionExponential? Actually I can see what it does. But,
why is it separated from the strategy itself?

RetxSuppression* implements an algorithm to determine whether an incoming
Interest should cause strategy retry or should be aggregated. Too much
retries waste bandwidth; too much aggregation limits the opportunity to
explore alternate paths https://redmine.named-data.net/issues/3889 .
The implementation is in separate classes so that it can be reused across
strategies. However, it should be considered part of the strategy design.


About you suggestion. Do we have any kind of implementation of BFD(rfc
5880/7880) for NDN in any level?

No.

Yours, Junxiao
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndnsim/attachments/20170220/88e16d43/attachment.html>