Hi, You can use sflow-rt to do this. The switches just need to support sflow protocol. we realized a module for floodlight able to measure the bandwidth and instantiate dinamically the flows changing the cost of the link. The code will be free and opensource but it hasn't be publicated officially yet.
anyone wants more infos can contact me by email at luca.pr...@garr.it Cheers, Luca > Il giorno 17/ott/2013, alle ore 20:34, Nicholas Bastin > <nick.bas...@gmail.com> ha scritto: > >> On Thu, Oct 17, 2013 at 11:19 AM, Alison Chan <chan7...@kettering.edu> wrote: >> Are there any papers which describe benchmarking bandwidth between >> switches in an OpenFlow-controlled network, or a mechanism within the >> OF spec to measure bandwidth? Topological discovery is easy enough, >> but I want to add weights to the graph edges which will be bandwidths. > > Do you want the *reported* bandwidth, or the actual measured (currently) > available bandwidth? Reported bandwidth can come from the usual places - > SNMP, etc. - and available bandwidth can be had using existing methods. The > fact that the switch is OpenFlow doesn't really impact your ability to do > available bandwidth estimations, other than the fact that per-hop test tools > will get screwed by the lack of end-to-end TTL decrement support in the stack > (even on switches that support and expose this, most controllers don't use > it). > > -- > Nick > _______________________________________________ > openflow-discuss mailing list > openflow-discuss@lists.stanford.edu > https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
_______________________________________________ openflow-discuss mailing list openflow-discuss@lists.stanford.edu https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
