[Nfd-dev] Best Route strategy doesn't work over ethernet multicast?

[Lan Wang (lanwang)]

Nick,

Best-route switches routes only in response to consumer retransmissions.  It does not measure the forwarding plane performance of the different faces and switch to the best one.  The best here means the best in routing cost (whatever routing ranks the best).  This is the reason why best-route won’t work when we have hyperbolic routing enabled, as hyperbolic routing may rank a suboptimal route (not the best in actual RTT) better than routes with shorter RTTs.   We cannot simply follow the best routing cost, so we introduced the ASF (adaptive SRTT-based forwarding) strategy which sends periodic probing interests to measure the RTT of routes with lower ranking.  You can find more details about ASF in our technical report:

An Experimental Investigation of Hyperbolic Routing with a Smart Forwarding Plane in NDN<https://named-data.net/publications/techreports/ndn-0042-1-asf/>.
Vince Lehman, Ashlesh Gawande, Rodrigo Aldecoa, Dmitri Krioukov, Beichuan Zhang, Lixia Zhang, and Lan Wang.
NDN, Technical Report NDN-0042.
https://named-data.net/publications/techreports/ndn-0042-1-asf/

Lan

On Mar 27, 2018, at 9:54 PM, Junxiao Shi <shijunxiao at email.arizona.edu<mailto:shijunxiao at email.arizona.edu>> wrote:

Hi Nick

I thought best-route would eventually switch if the one
it chose wasn't working, but it just kept blindly sending to the
numerically least-ordered face.

best-route v4 has no “memory” beyond a single PIT entry. It’ll use a secondary path only if the downstream retransmits before PIT entry expires. In another PIT entry, it still starts from the first nexthop.

best-route v4 does not deal with “numerically least-ordered face”. When two nexthops have the same cost in FIB, their order is unspecified.

Yours, Junxiao
_______________________________________________
Nfd-dev mailing list
Nfd-dev at lists.cs.ucla.edu<mailto:Nfd-dev at lists.cs.ucla.edu>
http://www.lists.cs.ucla.edu/mailman/listinfo/nfd-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/nfd-dev/attachments/20180328/aa63b408/attachment.html>