rkukura commented on a discussion. Hi Steven, The VLAN tags are being modified by OVS flow rules as packets cross the veth connecting the bridges. These flows can be displayed using ovs-ofctl as follows: ovs-ofctl dump-flows br-int NXST_FLOW reply (xid=0x4): cookie=0x0, duration=947602.247s, table=0, n_packets=150980, n_bytes=37605726, idle_age=1, hard_age=65534, priority=1 actions=NORMAL cookie=0x0, duration=947598.9s, table=0, n_packets=16652192, n_bytes=1367674080, idle_age=0, hard_age=65534, priority=2,in_port=2 actions=drop cookie=0x0, duration=110.96s, table=0, n_packets=4, n_bytes=1158, idle_age=48, priority=3,in_port=2,dl_vlan=1000 actions=mod_vlan_vid:2,NORMAL ovs-ofctl dump-flows br-eth2 NXST_FLOW reply (xid=0x4): cookie=0x0, duration=947604.202s, table=0, n_packets=17266871, n_bytes=1429653351, idle_age=0, hard_age=65534, priority=1 actions=NORMAL cookie=0x0, duration=947601.253s, table=0, n_packets=31, n_bytes=2318, idle_age=113, hard_age=65534, priority=2,in_port=4 actions=drop cookie=0x0, duration=113.882s, table=0, n_packets=107, n_bytes=6118, idle_age=1, priority=4,in_port=4,dl_vlan=2 actions=mod_vlan_vid:1000,NORMAL [root@rachel ~(keystone_bob)]# In this case, the local VLAN tag on br-int is 2, and the VLAN tag on the physical network (br-eth2) is 1000. Whether you need to enable VLAN splinters depends on what NIC driver you are using. I have not needed to run commands such as "ovs-vsctl set port eth2 trunks-81,82". We will be working to characterize any performance overhead from using multiple bridges connected by veths, and may explore a single-bridge solution if justified. Hope this helps, -Bob --- Follow the link below to check it out: http://openstack.redhat.com/forum/discussion/comment/660#Comment_660 Have a great day!