Quoting daniel marques (daniel.marq...@uniriotec.br): > Hi Serge, > > Sorry for my late answer. I do use veth, but the other experiments also use > it.
Please explain exactly how the non-container network namespace test was set up, though. > But my guess there is might be some kind of kernel customization, so > they can reach these speeds, but unfortunatelly I am not aware of what > customization they made on their kernels. > A little more about my setup... My kernel's version is: > 3.20.0-23-generica-pae. I don't know what this is, linux isn't at 3.20 yet :) Did you use the same kernel for both? > And the network setup of my containers: > lxc.network.type=veth > lxc.network.link=lxcbr0 > lxc.network.flags=up lxcbr0 is a NAT'd bridge so you'll have slowdown due to that. You might try specifying the same bridge you used for the netns tests. That would explain a difference in performance. -serge ------------------------------------------------------------------------------ Keep yourself connected to Go Parallel: TUNE You got it built. Now make it sing. Tune shows you how. http://goparallel.sourceforge.net _______________________________________________ Lxc-users mailing list Lxc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lxc-users
