IPerf Experts,

I am performing tests between an iPerf 3.1.3 client running on a Linux Ubuntu 
Virtual machine on a VMware ESX host and a similar setup with the iPerf 3.1.3 
server VM running on a separate ESX server. The ESX servers have 10 Gig NICs 
with plenty of CPU and both connect to a network with a 40 Gig core. I am able 
to drive 10 Gbps linerate bandwidth when running iPerf "TCP" flows between the 
client and server. However, the most I can get when running a "UDP" test is 
roughly 5.5 Gbps transmission rate. I actually tweaked a couple of iPerf 
parameters to increase the transmission rate that ultimately got me to 5.5 Gbps 
-- but that is all I can get so far. It doesn't look like the VMs running iPerf 
are CPU bound so it must be the iPerf UDP settings are not optimized.

Can someone provide guidance on the optimal iPerf settings to get the maximum 
transmission rate when using a UDP flow ??

Side question: When running iPerf 3.1.3, I was not able use the "-u" flag on 
the server as it stated this was an invalid argument. I also tried changing the 
listening port as well with no luck. I was able to have the client set the UDP 
test and the server was able to accept the UDP test from the client. I just 
can't set the options on the server side of iPerf 3.1.3 the way I can with the 
2.x releases. I'm not sure if this is something on my end or if this is 
expected behavior. I also need to change the port the server listens on and 
can't get that working when using the "-p" flag on the server nor when I try to 
set the server side port on the client.

Sorry, just a few more questions. I see there is a way to set the TOS with the 
"-S" flag. Is there a way to set the DSCP (Diffserv) marking in the packet 
instead of TOS ?? Lastly, is there a way to set the MTU size of the frame 
generated by iPerf or possibly even use jumbo frames ??

Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Iperf-users mailing list

Reply via email to