[lxc-users] Has anyone looked at an LXC solution like the proposal for native Docker Multi-Host Networking

[brian mullan]

Just curious if anyone knows of any work going on around something like
this Docker proposal for Multi-Host networking:

https://github.com/docker/docker/issues/8951

Given the direction of LXD/LXC and their increased use in data centers &
clouds I would like to know if anyone has looked at the proposal for native
Docker multi-host networking in regards to an LXD/LXC solution for the same
problem(s).

I've been looking at VXFLD (vxlan flood)
<https://github.com/CumulusNetworks/vxfld> which was open-sourced last year
by Cumulus Networks and MetaCloud (just before Cisco bought MetaCloud).

VXFLD attempts to solve the problem handling BUM (Broadcast, Unknown &
Multicast) packets for VxLAN without requiring that the network support
Multi-Cast.

VXFLD was submitted to IETF as a proposed uni-cast VxLAN solution but I'm
not sure where it stands.

re. a uni-cast VxLAN solution which would enable Overlay networking of LXC
containers.

These 2 topics (multi-host networking) and what VXFLD attempts to solve
(unicast VxLAN) are big topics in OpenStack & SDN today.

I've been thinking that either an LXC "host" could be used as the VxLAN
VTEP (Virtual Tunnel End Point) and traffic from LXC containers on that
host could be inter-connected with any other LXC container on any other
Host via the VxLAN tunnel.

Just thought I'd ask if anyone else was or has looked at this kind of
problem/solution in LXC deployments?

Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxcontainers.org/pipermail/lxc-users/attachments/20150117/9e48de80/attachment.html>