[alcatel-nsp] PBB and IGMP Snooping

Thedin Guruge thedin at gmail.com
Wed Jan 29 22:56:53 EST 2014


Hi Jorge,

Thanks for you response, please keep me in-line,

Regards,

Thedin


On Wed, Jan 29, 2014 at 7:24 AM, Rabadan, Jorge (Jorge) <
jorge.rabadan at alcatel-lucent.com> wrote:

>  Hi Thedin,
>
>   From: Thedin Guruge <thedin at gmail.com>
> Date: Tuesday, January 28, 2014 at 2:17 AM
> To: "alcatel-nsp at puck.nether.net" <alcatel-nsp at puck.nether.net>
> Subject: [alcatel-nsp] PBB and IGMP Snooping
>
>    Hi All,
>
>  when IGMP snooping is enabled on an i-vpls instance, is it possible to
> activate 'send queries' towards the b-vpls?
>
>
>  'send-queries'  is configured under saps and sdp-bindings. On the i-vpls
> you can't active send-queries to be sent to the b-vpls.
> You will need a real querier.
>
> [TG] Thanks for confirming, w.r.t our setup we could rely on an upstream
querier when 'b-sdp' where the MFIB entry is pointed to gets changed due to
far-end failure, I've noted that in such case, the previous MFIB will still
remain until aged out, but new MFIB entries get created when reports are
received on the new sdp due to queries, until such time, traffic will be
flooded.

>
>  Also, when b-vpls's FDB state changes, does it propagate to i-vpls MFIB?
> i.e. If shared mac (mc-lag) to sdp mappings on b-vpls changes (via
> mac-notification), can that propagate into to the i-vpls MFIB??
>
>
>  Mac-notification sends a message with the shared sap-bmac as the source
> address from the new active PE, so that the remote PEs can update their
> FDBs. This mechanism allows to avoid black-holes for unicast traffic. For
> multicast traffic there is no blackhole since the traffic is flooded into
> the (*,G) tree... do you have any issue?
>
> [TG] Yep i understand, the intent we had here was to reduce flooding when
MFIB entry becomes invalid due to far-end PE mc-lag failover, i.e. when
b-vpls FDB is updated if that could trigger the MFIB to be updated as well,
then the temporary flooding (until MFIB is aged out) could be prevented.
Agreed flooding is better than blackholing, but with increased no. of
services, flooding traffic volumes can be substantial.

>
>
>  --
> Regards,
>
>  Thedin Guruge
>
>


-- 
Regards,

Thedin Guruge
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/alcatel-nsp/attachments/20140130/4ab7e809/attachment.html>


More information about the alcatel-nsp mailing list