Hi :
At the meeting there was the question about the scalability that the backbone router provides. I can add a few words on that. I'd suggest text along the following lines: - In some cases, the scalability we are looking for is simply geographical, extending the LoWPAN over a large space. The backbone can interconnect localized LLNs into a larger one. - In other cases, we want to allow more nodes in the network, and avoid the need to notify all LBRs of all addresses by distributing the white board over the backbone. The number of nodes that the Backbone router allows is basically limited by the capability of the backbone, which varies a lot between a WIFI mesh and a 10+Ge TRILL infrastructure. What we can say on the positive side is that: - The traffic usually converges to a limited set of routers on the backbone. Those will have to manage large neighbor caches. But considering the type of traffic, most other nodes on the backbone will have small caches, and that illustrates that the traditional reactive ND over the backbone is more efficient for our needs than a proactive routing protocols that would force all BBRs to know all devices in the networks. - The backbone routers can throttle the ND traffic for their devices so the traffic will be smoother than with a same number of devices really connected to the backbone. - The devices are battery operated so you do not have an effect like a brutal power on of all nodes. Do I miss anything? Pascal
_______________________________________________ 6lowpan mailing list [email protected] https://www.ietf.org/mailman/listinfo/6lowpan
