On 06/05/14 14:09, govardhan sankanna wrote: > I am planning to use netmap-rumptcpip to increase the performance. So, I > was trying to measure the PPS and CPS by writing a small client server > program. > > But, unfortunately I am getting very small CPS, it is coming in the range > of few hundreds where as if I use without this I get in thousands. So, I am > thinking I might be doing something wrong in using netmap-rumptcpip. Would > it be possible for you to have a look at it and advice why I am getting > such a low rate. > > I have modified the netmapcat program to measure the number of tcp > connections it can accept per second. I have attached the program.
Your program is not really measuring CPS that the networking stacks can handle. You should modify the client side to use O_NONBLOCK to open multiple connections simultaneously and poll for their completion. That said, it might be simpler to use an existing benchmarking package which makes things like the above into account. One possible starting point is netperf: http://wiki.rumpkernel.org/Howto%3A-netperf ------------------------------------------------------------------------------ Is your legacy SCM system holding you back? Join Perforce May 7 to find out: • 3 signs your SCM is hindering your productivity • Requirements for releasing software faster • Expert tips and advice for migrating your SCM now http://p.sf.net/sfu/perforce _______________________________________________ rumpkernel-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/rumpkernel-users
