Shared memory version of tap+vhost_net to support a userspace switch?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hello,

I am writing on behalf of a newly developing open source project
called Snabb Switch (http://snabb.co/snabbswitch/). This is a
Lua-scriptable userspace vswitch that's designed for performance, as
funny as that may sound at first :-). The goal is to have an
über-extensible ethernet fabric that runs on any kernel and can be
used to implement the wildest of network virtualization fantasies.

The project has a technical goal of supporting zero-copy I/O both
towards physical NICs and virtual NICs in KVM guests. The physical
side is up and running now
(http://blog.lukego.com/blog/2013/01/05/kernel-bypass-networking-in-snabb-switch/)
and it's time to start working on the virtual KVM side.

So! We need a way for the userspace switch process to talk directly
with a KVM guest. That sounds like an equivalent of tap+vhost_net that
connects the guest's virtio driver directly with another userspace
process instead of with the host kernel (as tap+vhost_net does today).

I can take care of the switch side of this virtio interface but I
would absolutely love help with the KVM side. Is anybody interested in
hacking this or otherwise contributing towards an implementation? It
would be a huge contribution to the project.

And thank you for reading this far :-) and many thanks to Stefan
Hajnocz for his patient help already!

Cheers,
-Luke
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[Index of Archives]     [KVM ARM]     [KVM ia64]     [KVM ppc]     [Virtualization Tools]     [Spice Development]     [Libvirt]     [Libvirt Users]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite Questions]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux