Re: Draft charter for L2VPN: Perfect LAN emulation

Alex,

Alex Zinin wrote:
> 
> Reading the discussion between Matt, Eric, et al, it seems that
> we have two main questions at hand:
> 
>  1. Should the charter define the level of transparency of
>     the L2 VPN mechanisms to the higher-layer protocols, and
>     if so, what should the required level be?
I think a definition of the service being specified in the WG would be
needed. 

Wrt level of transparency, I think the minimal requirement is any
non-transparent characteristics must not affect the operation of these
higher layer protocols.

For VPLS, a definition of the service being specified in the WG would be
helpful. 
Here's one suggestion :
An emulated LAN service allows the inter-connection of (geographically)
dispersed devices, not connected over a LAN segment, as if they were
attached to a single LAN.
Hence standard compliant devices (e.g. bridges, routers, hosts) would be
able to work as if they are connected to a LAN when connected to the
emulated LAN service.

An emulated LAN service operates below the MAC Service Boundary, and is
transparent to protocols operating above this boundary.
The emulation may be imperfect in terms of QoS (limitation should be
documented in solution specifications), compared to operation over a
LAN, as long as the differences/degradation in the emulated service does
not prevent standard compliant devices from functioning. However, the
emulated LAN service must not introduce artifacts which break the
operation of standard compliant devices.

>  2. Should the charter say that the WG will work with IEEE 802.1
>     to ensure proper interworking.
> 
> My reading of the discussion on the first item is that we don't have a
> strong case for spelling out that a perfect emulation is required.
> If anyone disagrees, please speak up.
I think a definition of the desired LAN emulation service is useful. I
am not clear if there is agreement on this or what is "perfect
emulation".

> I didn't see a discussion on the second issue, but it seems like a
> reasonable proposal to me.
Yes.

Thanks
Cheng-Yin