[ovs-discuss] ovs and vxlan tunnels

Scott Lowe scott.lowe at scottlowe.org
Thu Aug 4 16:59:51 UTC 2016


Please see my response below.


> On Aug 3, 2016, at 8:52 PM, Yinon <yinonby.hpe at yahoo.com> wrote:
> 
> Hello,
> 
> I have a question about ovs and vxlan tunnels.
> I am interested in ovs as a vswitch in order to achieve the following topology:
> I have 2 nodes, each with 1 NIC interface, say eth0. I would like to have  2 networks (subnets) without using vlans, say 10.100.1.0/24 and 10.200.1.0/24.
> 
> So something like a bridge, say br-ex, with eth0 added to it.
> And 2 internal-type ports on that bridge, say port-100 and port-200. port-100 is assigned an IP from subnet 10.100.1.0/24, and port-200 from 10.200.1.0/24.
> Traffic that enters the bridge via eth0 on vxlan 100 will be decapsulated and forwarded to port-100, and traffic that enters on vxlan 200 will be decapsulated and forwarded to port-200. in the opposite direction vxlan will be encapsulated accordingly.
> 
> I can probably do it with openflow rules, but was wondering if there is a more elegant way, using vxlan-type ports?


Have a look at the OVS articles on my web site (<http://blog.scottlowe.org/tags/#OVS>). I have some posts there on doing exactly what you're describing using GRE, but the concepts should be the same for VXLAN. Sorry I can't give you a more specific link; standing in line to board an airplane at the moment. :-)

-- 
Scott

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20160804/38e0f089/attachment-0002.html>


More information about the discuss mailing list