clarification of Layer 2 sub-layers

Hello Himanshu,

Nice to hear from you again.

My understanding of the below discussion is that Ali and Mick are agreed
on the following:

- Spanning Tree will not run on service providers networks natively
- Islands of Spanning Tree may exist on service provider networks
- Spanning Tree BPDUs may be encapsulated across PW connections

I see the layer model in the following way:

Higher layer (usually IP)
VLAN (optional)
Spanning Tree
Encapsulated Ethernet (only on the P-P or P-PE connections)
MPLS (only on the P-P and P-PE connections)
Ethernet or core network equivalent

Does everyone agree that this is the correct order of the layer model?
ST resides below VLANs because BPDUs are never tagged (even in the case of
802.1s).

Maybe this should be clarified in the proper drafts so that everyone is
clear.

Thanks,
-Ben

On Mon, 12 May 2003, Mick Seaman wrote:

> Himanshu,
>
> I don't see the disconnect between what Ali is saying below and what I said.
> They may be different statements but I don't see that that means they
> disagree. Unfortunately I don't see how to resolve your difficulties. Mick
>
> -----Original Message-----
> From: Himanshu Shah [mailto:[email protected]]
> Sent: Monday, May 12, 2003 2:43 PM
> To: Ali Sajassi; [email protected]; Norman Finn; Muneyoshi Suzuki;
> Tony Jeffree; Les Bell; Paul Congdon; Neil Jarvis
> Cc: [email protected]; [email protected]
> Subject: RE: Comments on the L2 VPN framework and solutions documents
>
>
> Ali,
>
> You seem to have different opinion than others
> in IEEE group.
>
> ali> I don't think there is a disconnect. When you talk about
> ali> P-STP, you need to consider both intra-island and
> ali> inter-island. P-STP is NOT used for inter-island -
> ali> for that we use full-mesh with split horizon.
> ali> Also keep in mind that P-STP is suggested for
> ali> Ethernet access domain and not MPLS access domain.
>
> However, Mick seems to suggest something different
>
> mick> If the VPLS drafts have not changed significantly in
> mick> their intent (I don't think they have) there is not a disconnect here.
>
> mick> There is a difference between running spanning tree to cut PWs
> mick> and reduce their topology to that of a spanning tree
> mick> (which is what the VPLS folks were correctly rejecting)
> mick> and conveying spanning tree BPDUs across the emulated LAN
> mick> formed by PWs so that bridge ports attaching to that emulate LAN
> mick> service can have the right state, including the 'LAN' or not in the
> mick> active topology/topologies for VLANs. Again it is neccessary to get
> mick> the layering straight, focusing on which wire a BPDU goes down before
> mick> the layering is straight really doesn't help.
>
> Can you clarify what you mean?
>
> /himanshu
>