[ndnSIM] RIB do not work well
shijunxiao at email.arizona.edu
Thu Apr 26 03:50:09 PDT 2018
> In this scenario, there are three different routes to destination. one
> route with 3 hops to destination and 1 mbps data rate in bottleneck link,
> another with 4 hops and 1 mbps data rate in bottleneck link and the third
> one with 4 hops but 10 mbps data rate in bottleneck link.
# srcNode dstNode bandwidth metric delay queue
> Src1 Rtr1 10Mbps 1 10ms 20
> Rtr1 Rtr2 10Mbps 1 10ms 20
> Rtr1 Rtr4 1Mbps 1 10ms 20
> Rtr3 Rtr4 10Mbps 1 10ms 20
> Rtr1 Rtr3 1Mbps 1 10ms 20
> Rtr2 Rtr4 10Mbps 1 10ms 20
> Dst1 Rtr4 10Mbps 1 10ms 20
Despite having different bandwidth, your "metric" setting did not reflect
that, so there's no way for RIB to tell their differences.
Both multicast and best route v4 strategies do not attempt to detect link
bandwidth or Interest satisfaction rate. Their decisions are solely based
on metric setting.
To fix this problem, set the metric to the inverse of bandwidth.
in addition I found that in multicast strategy, interests do not multicast.
> In fact, it works as the best strategy works.
> ndn::GlobalRoutingHelper ndnGlobalRoutingHelper;
The GlobalRoutingHelper only installs a shortest route from every node to a
prefix. Even if you are using multicast strategy, the Interest goes on only
one path because there's only one nexthop in the FIB entry.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the ndnSIM