Thanks for the reply. The Y-axis shows the number of times a delay value on X-axis occurs.
The router which I am using is Mikrotik RouterBoard 750GL, and I have erased the factory installed firmware. As a replacement, I have installed for OpenWRT ‘Chaos Calmer’ with Open vSwitch 2.1.2 (OvS) package. I have configured OVS by adding a bridge, ports to it and controller through ‘ovs-vsctl’ command. I am using Distributed Internet Traffic Generator for generating traffic. I have repeated the experiments on GENI using OVS as an OpenFlow switch configured as explained in the this tutorial http://groups.geni.net/geni/wiki/GENIExperimenter/Tutorials/OpenFlowOVS/DesignSetup . The same set of experiments are repeated on Mininet using OVS switch. The results of all the experiments are same that as the rate increases overall delay decreases. I am using OVS in all the three platforms. Any specific reason for decrease in delay? Does it relate to how OVS is processing the flows and can I verify it somehow? Uzzam On Sat, Dec 19, 2015 at 4:26 AM, Justin Pettit <[email protected]> wrote: > I'm not sure I understand the Y-axis. Regardless, it sounds like you're > not using a hardware OpenFlow switch, and not OVS, so you're probably > better off asking them. I'd guess it has to do with the flows being > cached, but it's complete speculation, since it's going to depend on how > they implemented their OpenFlow stack. > > --Justin > > > > On Dec 17, 2015, at 10:16 AM, Uzzam Javed <[email protected]> > wrote: > > > > Hi everyone, I am doing some experiments on a commercial router enabled > with OVS. I have connected two hosts to two ports of the router and have > added both ways flow entries through ‘ofctl’ command. I am generating round > trip traffic through a traffic generator at a constant rate. The total > number packets are kept constant to 10,000 packets. When I increase the > rate the overall delay decreases. The attached picture shows the density > plots of the 3 experiments having rate 10, 100 and 1000 packets/s input > traffic rate, respectively. > > > > > > > > What is the reason for these results that why delay decreases with > increase in input traffic rate? > > > > > > > > Thanks, > > > > Uzzam > > > > <rate_incr.jpg>_______________________________________________ > > discuss mailing list > > [email protected] > > http://openvswitch.org/mailman/listinfo/discuss > >
_______________________________________________ discuss mailing list [email protected] http://openvswitch.org/mailman/listinfo/discuss
