[ovs-discuss] Identical VM-MACs with Open vSwitch?

Ivan Pepelnjak ipepelnjak at gmail.com
Sat Dec 10 04:37:18 PST 2011


According to 802.1Q, the forwarding decision could be made based on dMAC+VLANtag. For those that want to have standard citations ;) ... 802.1Q-2005, section 7.2 para (b)

"Effectively extend the source and destination MAC addresses, by treating frames and addressing information for different VLANs independently;"

More precisely, 802.1Q defines "Shared VLAN learning" (which seems to be what vSwitches are doing based on this discussion) and "Independent VLAN learning" (which is what this discussion seems to be all about). 

Most physical switches use "Independent VLAN learning" and the networking engineers are used to that behavior. It's also a nice-to-have behavior as it allows you to implement HSRP/VRRP in L3 VM appliances without having to consider global HSRP/VRRP group numbering.

Just my $0.02, you need more, you know where to find me ;)
Ivan

> -----Original Message-----
> From: discuss-bounces at openvswitch.org [mailto:discuss-
> bounces at openvswitch.org] On Behalf Of Trebor Forban
> Sent: Saturday, December 10, 2011 1:15 PM
> To: discuss at openvswitch.org
> Subject: [ovs-discuss] Identical VM-MACs with Open vSwitch?
> 
> Hello Ben,
> 
> thanks for your answer. I'm not really too familiar with VLANs, but I
> did look into this once with vde-switch from virtualsquare.org, and
> identical MACs were not supported. I suspect there might be VLAN
> switch implementations where uniqueness/routing is determined by
> MAC+VLAN bits, but I'm only guessing. I will look into this and
> OpenFlow, but it would be nice if someone who is in the know could
> lead me in the right direction, before I spend endless hours barking
> up the wrong tree.
> 
> Best regards,
> TF
> 
> 2011/12/9 Ben Pfaff <blp at nicira.com>:
> > On Fri, Dec 09, 2011 at 11:05:16AM +0100, Trebor Forban wrote:
> >> I would be very thankful for any insight, as to how I might be able to
> >> "cloak" identical MACs.
> >> Can Open vSwitch realize this with VLANs, or maybe some type of MAT
> (MAC-NAT)?
> >
> > Open vSwitch supports VLANs, so if VLANs solve your problem then yes
> > that is one possibility.
> >
> > I think that you could implement MAC translation using OpenFlow
> > rules.  I have not implemented a MAC translation solution before, so I
> > don't know what issues are likely to come up.
> _______________________________________________
> discuss mailing list
> discuss at openvswitch.org
> http://openvswitch.org/mailman/listinfo/discuss



More information about the discuss mailing list