I want to send stacked vlan frames to my ESX host from my switched network in a carrier / service provider environment. These will be outertag (ethertype 9100) : inner tag (ethertype 8100). I'd like the vswitch to deal with the outertag, and pass the inner frames , as tagged frames , to the guest.
example:
1) physical switch network sends frame as tag 3000 (type 9100) : tag 100(type 8100) to esxi host.
2) esxi vswitch is configured with vlan 3000 (type 9100) so it pops that tag and...
3) forwards the frames as tag 100 (type 8100) to the guest.
4) 8100 is the default ethertype, so we just configure the guest with a vlan tag of 100 , most guest OS can do this, and it should work.
But..
how on earth can I change the ethertype of the tags that the vswitch works with? I have access to both ESXi old-style (in lab) and Nexus (in production) v-switching systems.
I have read one or two workaround suggestions in this forum:
- I don't want to pass the full q-in-q stack to the guests (i.e. setting vlan 4095 VGT mode on the vswitch and passing all through) as q-in-q support on all guest OS that you may ever need to use is very unlikely.
- I don't feel keen to install some linux guest OS configured a bridging device and pass all through packets through that on the way to real guest (pushing / popping tags as appropriate), in a large service provider network design this doesn't tick the "scalable , resilient, stable" type of boxes.
suggestions?
Ian
