Dear,

Thanks you for your reply. Let me recall back my wireless topology
Gateway1(RYU Controller)--------Node 1-------Node 2--------Gateway2.
 Iperf -s command at Gateway 2 (default window size: 85.3 KByte) and iperf
-c <ip address of Gateway 2> -t 86400 -i 10 at Node 2. The testing for
iperf between Node 2 and Gateway 2 could be run for 24 hours.
Running iperf -s command at Gateway1(default window size: 85.3 Kbyte) and
iperf -c <ip address of Gateway1> -t 86400 -i 10 at Node 1. The testing of
iperf between Node 2 and Gateway 2 could be only run 200 seconds. Then
increasing window size up to 195 Kbyte at Gateway 1 increase the running
time up to 10000 seconds for TCP iperf experiment between Gateway 1 and
Node 1. I will share the experience for the root cause after the
investigation has been done.

Best Regards,
Soe Ye Htet

On Wed, Sep 19, 2018 at 2:39 PM IWAMOTO Toshihiro <iwam...@valinux.co.jp>
wrote:

> On Tue, 18 Sep 2018 21:07:41 +0900,
> Soe Ye Htet wrote:
> >
> > Dear,
> >
> > Please let me extend my case in this problem. The real wireless network
> > topology I have applied is Gateway1------Node 1------Node
> > 2--------Gateway2. RYU controller is installed in Gateway 1. The control
> > type is in-band control type by sharing Control TCP packet and Data TCP
> > packet on the same physical wireless interface. Then the experiment has
> > been executed to investigate about the TCP wireless link throughput. At
> the
> > side of Gateway1, iperf -s is executed. At the side of Node 1, iperf -c
> <ip
> > address of gateway1> -t 10000 -i 10. At the side of Gateway2, iperf -s is
> > executed. At the side of Node 2, iperf -c <ip address of gateway2> -t
> 10000
> > -i 10. Strangely, the iperf testing between Gateway 1 and Node 1 is
> stopped
> > at the time of 1000 seconds while the iperf testing between Gateway 2 and
> > Node 2 can be executed up 10000 seconds. I define the same flows for
> iperf
> > testing at the gateway 1 and gateway 2. For eg, i define the flow rules
> for
> > TCP at gateway 1 to receive the TCP packet from node 1. I also define the
> > same way of flow rules for TCP at gateway 2 to receive the TCP packet
> from
> > node 2. If anyone face same kind of issue, please let me know your
> > experience in solving for this problem.
>
> You seem to have already asked at the iperf mailing list.
>
> Unless you changed the flow tables during the experiment, I don't
> think ovs is an issue. You might be able to confirm this by having a
> table-miss flow entry and checking matched packet counts.
>
> I guess more likely cause is radio interference. Somebody may have
> used a microwave oven or other WiFi communcitation.
>
> > Best Regards,
> > Soe Ye Htet
> > Future Internet and Wireless Research Group
> > Chulalongkorn University
> >
> > On Sun, Sep 16, 2018 at 8:15 PM Soe Ye Htet <soeyehtet1...@gmail.com>
> wrote:
> >
> > > Dear RYU Team,
> > >
> > > I am trying to test the TCP link throughput between RYU controller and
> OF
> > > Switch in my software defined wireless network. The implementation of
> > > control plane is in-band by sharing the control traffic and the data
> > > traffic in same physical wireless link.  I plan to execute TCP
> throughput
> > > between RYU controller and one OF switch by executing iperf command.
> At the
> > > side of RYU controller, the command line of " iperf3 -s -d" is
> executed and
> > > the command line of " iperf3 -c <ip address of RYU> -t 10000 -i 10" at
> the
> > > OF Switch. Iperf3 was be operated up to 9600 seconds and then the TCP
> > > connection between RYU controller and OF Switch. I am trying to
> investigate
> > > what is real root cause of this problem. Can it be because of the
> failure
> > > of TCP handshake between RYU controller and the OF switch.
> > >
> > > Best Regards,
> > > Soe Ye Htet
> > > Future Internet and Wireless Research Group
> > > Chulalongkorn University, Thailand
> > >
>
> --
> IWAMOTO Toshihiro
>
_______________________________________________
Ryu-devel mailing list
Ryu-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ryu-devel

Reply via email to