Open Stack

On 21 June 2014 17:17, A, Keshava <keshava.a at hp.com> wrote:

> Hi Thomas,
>
> This is interesting.
> I have some of the basic question about deployment model of using this
> BaGPipe BGP in virtual cloud network.
>
> 1. We want MPLS to start right from compute node as part Tennant traffic ?
> 2. We want L3 VRF separation right on Compute nodes (or NN Node) ?
>         Tenant = VRF ?
>         Tenant span can be across multiple CN nodes,  then have BGP to
> Full mesh with in CN ?
> 3. How to have  E-VPN connectivity mapping at NN/CN nodes ?
>     Is there an L2 VPN psuedowire thinking from CN nodes itself ?
> 4. Tennant traffic is L2 or L3 or MPLS ? Where will be L2 terminated ?ic
> Routing bp.
>
>
When you say things like 'tenant = VRF' (and, in fact, I presume you mean
'network = VRF', since networks and tenants are two different things) then
that's actually more to do with how you implement the networking overlay
layer in Neutron.  While interesting, and while it could potentially use a
BGP speaker and VRFs to do it, it's not the same use case as advertising
routes externally, or terminating an external MPLS VPN in Openstack.  I
could do either of those things independently of the other.  Terminating
VPNs requires an extension API and could potentially glue on to a current
plugin (much like VPNaaS).  Writing overlays is a new plugin entirely.

There's a use for BGP speakers in both arenas (and additionally within the
distributed virtual router, which is a third case) so perhaps we could
address who's using which speaker for precisely what?
-- 
Ian.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140622/30f1e55e/attachment.html>