On 05/10/10 01:24 AM, Pascal Thubert (pthubert) wrote:
Hi Erik:
OSPFv3 has similar instances.
How does a host know which "instance" to use in the case of OSPFv3?
Why doesn't ROLL use the same mechanism?
Traffic Engineering is all about mapping flows of traffic onto complex
topologies that are designed for those flows.
So nothing real new. RPL just makes that a bit less manual, because we
have harder constraints in terms of scalability and so-called
autonomicity.
RPL dynamically builds its TE topologies and specifies the flow label
interaction a bit deeper to automate the mapping of the flows onto those
topologies.
I see that as extension or evolution, but not revolution. A survival
trait for the Internet Architecture, really.
If there is a need to evolve the Internet Architecture, it is best not
done in a narrowly focused IETF working group.
I suggest that be discussed in the INT-AREA WG to begin with.
But fundamentally I don't understand why a working group like ROLL that
seems keen on building small and simple networking devices at the same
time feel a requirement to add bells and whistles like this. Seems to be
counter to designing a simple and robust network infrastructure.
Erik
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan