RE: [RPRWG] A question on IP over RPR
> This is responsibility of the upper layers. There are several ways this
can
> be done. One approach would be to recognize each ringlet as a separate
> interface. Each interface would then show up as a routable interface and
if
> the same route is learnt from both these ringlets, the routing protocols
> will make a choice based on configured weights, costs and other parameters
> to choose "least cost forwarding" interface and installs that route in the
> kernel routing table.
>
> The other approach involves exporting only one RPR interface for all the
> ringlets to the routing protocols. The decision on how to forward the data
> is something that can be done at the RPR MAC client level because the
> topology information is present in the MAC.
>
> In other words, either approach is supportable by RPR and the decision
> should be left to individual network configurations and the user.
Right -- either approach can be done, and I think the choice doesn't have to
be made by 802.17. The choice between these two approaches is probably a
good topic for the IETF IPORPR WG, which I understand is mostly inactive
awaiting 802.17 WG progress.
-- Jim
> -----Original Message-----
> From: lcwang@xxxxxxx [mailto:lcwang@xxxxxxx]
> Sent: Thursday, January 31, 2002 1:12 AM
> To: stds-802-17@xxxxxxxx
> Subject: [RPRWG] A question on IP over RPR
> Importance: High
>
> A question of IP over RPR.
>
> According to IP address, using routing protocol, a data forwarding route
can
> be selected. On the other hand, RPR may transfer data through
> its ring on both direction. How can consistency of L3 and L2 data transfer
> be guaranteed?
>
> Can someone explian this? Thanks.
>
> Leo Wang
>
>