Am 20.02.2013 11:06, schrieb Ray Hunter:
given that border routers may or may not advertise a default route, and may
also advertise
more specifics...
True. My assumption is that the 99% use case will be a simple default
route to "the Internet." That could be incorrect.
What's the use case behind more specific routes (in the walled garden)?
What would be so bad in the case of a walled garden if we always
installed a default SADR route associated with the prefix learned
specifically from the walled garden i.e.
(walled_garden_derived_prefix/56,::/0) -> NH_walled_garden_homenet_BR?
True, generally it seems to me, that more specific routes are only
needed in the case the Source Prefix in SADR is *.
Otherwise the route is constrained by the source address anyway.
The walled garden route is needed though, if the table looks like this:
(*, ::/0) -> ISP_A # Default route to ISP A
(*, ::/0) -> ISP_B # Default route to ISP B
(*, 2001:db8::/64) -> R1 # Internal network, prefix from A
(*, 2001:db8:1::/64) -> R2 # Internal network, prefix from A
(*, 2001:db9::/64) -> R1 # Internal network, prefix from B
(*, 2001:db9:1::/64) -> R2 # Internal network, prefix from B
(*, fd00::/64) -> R3 # Internal network ULA
(2001:db9::/56, 2001:420::/32) -> ISP_B # Walled garden route from ISP B
Maybe that should be listed as an example as well. And maybe a line
about how to behave if there's a tie with the multiple default route
constellation:
(*, ::/0) -> ISP_A # Default route to ISP A
(*, ::/0) -> ISP_B # Default route to ISP B
If we don't propose a solution , we should add a line which says that
it's up to the implementation or load balancing mechanism or whatever to
solve that, like:
A router forwarding a packet does a longest match look-up on the
destination address. If this is a (*, D) entry, it forwards the
packet out the best next-hop as before (doing equal cost multi path
load balancing etc). This applies also to multiple (*, ::/0) entries.
We've anyway still got to sort out correct source address selection even
once we've got the basic packet routing in place. Source address
selection could be the point where the real "routing" decision is
made, whilst SADR would be used to select the correct Homenet BR and to
avoid the ingress filters once the correct source address has been decided.
The hosts could perform NAT upon receipt of the ICMP code 5. Of course,
if you have a low power device then you might have a problem... just a
thought.
I like the draft and find it very useful.
_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet
