Morgan in the current ZC implementation you do not have to change anything in the guest or patch QEMU, but just load a kernel module (see https://svn.ntop.org/svn/ntop/trunk/PF_RING/userland/examples_zc/README.kvm <https://svn.ntop.org/svn/ntop/trunk/PF_RING/userland/examples_zc/README.kvm>).
The us-vhost approach is interesting and we might consider supporting it. But not in a month or two, as we’re actively working at other projects (see our blog) that we will start introducing this spring. Regards Luca the enhancement > On 11 Mar 2015, at 06:32, Morgan Yang <[email protected]> wrote: > > Hi: > > Recently the Intel DPDK guys presented about a "us-vhost" approach, which > they added DPDK support to KVM's side of vhost. See presentation here > http://openvswitch.org/support/ovscon2014/18/1530-dpdk-accelerating.pptx > <http://openvswitch.org/support/ovscon2014/18/1530-dpdk-accelerating.pptx> > > The nice thing about this is approach is least intrusive to the guest. For > "black box" or bundled virtual appliance deployments, we don't have the > ability to change the guest's virtio module, so we don't be able to integrate > with PF_RING_ZC and Netmap VALE. > > I'm curious if PF_RING is planning on something similar. > > > Much Thanks > Morgan Yang > _______________________________________________ > Ntop-misc mailing list > [email protected] > http://listgateway.unipi.it/mailman/listinfo/ntop-misc
_______________________________________________ Ntop-misc mailing list [email protected] http://listgateway.unipi.it/mailman/listinfo/ntop-misc
