RE: draft-kompella-l2vpn-vpls-multihoming-01

["HENDERICKX Wim" <[e-mail]>]

Bhupesh,

Thanks, In this case does the solution allows 2 different designated
forwarder for 2 different VE(s) on the same PE as per the example below.
I would assume it does. PE3 would have to maintain a PW to each PE (PE1
and PE2 in the example).

As you mentioned below we should extend the solution with a MAC flush
capability like LDP does today.

Cheers,
Wim 

-----Original Message-----
From: Bhupesh Kothari [mailto:[e-mail]] 
Sent: maandag 4 augustus 2008 20:51
To: HENDERICKX Wim
Cc: [e-mail]
Subject: Re: draft-kompella-l2vpn-vpls-multihoming-01

HENDERICKX Wim <[e-mail]> wrote:

> During the session I expressed concerns regarding 2 dual homes CE(s) 
> on 2 PE(s) with multi-homing. Here is the scenario I had in mind.
> 
>                             (VE1, VE2)
>                 ____ PE1....................
>               /  ____/  :                               :
>         __/   /         :                                PE3
> CE1 __   /          :                                 :
>             \ /             :                              :
>              / \_____ PE2 ..................
>        __/               /  (VE1, VE2)
> CE2 _________/
> 

> PE3 picks a designated forwarder lets say PE1 for both CE1 and CE2. 
> After discussion what I understood is that when CE1 fails PE3 reelects

> the designated forwarder for both CE1 and CE2 and PE3 also flushes the

> Mac table.

When CE1--PE1 link fails, all PEs will reelect who is the designated PE
for VE1 (assuming VE1 is for CE1).  Since the designated PE in this case
will move from PE1 to PE2, PE3 needs to flush the MACs learned from PE1.

> 
> This should be addressed as an impact on CE1 should not impact other
> CE(s) according to me.
> 

In VPLS, there is only one PW between a pair of PEs as the ACs are not
tied to particular remote neighbors as in the case of P2P L2VPNs.
Since PWs are not tied to ACs, when a AC goes down there is no way to
flush just the MACs associated with a particular AC other than
explicitly sending a list of MACs to be flushed.  In your example, PE1
can send a list of MACs (learned from CE1) to be flushed to PE3.  The
capability to send a list of MACs to be flushed is not yet there, but
can be added.


> Cheers,
> Wim


Thanks
Bhupesh