6bone connectivity

[Bob Fink LBNL]

I agree with Dimitry's suggestion.  Let's try it on for size and see how it
works!


Bob

==============================================
At 9:00 AM -0800 2/3/97, Dimitry Haskin wrote:
>Falks,
>
>We're currently experience a very pure 6bone connectivity.
>I think that the current situation is quite detrimental
>to 6bone effort. IMO, the main problem is that some/all core
>routers still use static routing to route between themselves.
>This leads to routing loops as well as to blackholing traffic to
>disabled routers even if an alternative path is available.
>
>To improve 6bone routing I propose to adopt the following
>policies:
>
>- restrict the backbone sites to only those routers that
>  support RIPng;
>
>- make the backbone routers to use exclusively RIPng to
>  route between themselves;
>
>- the backbone routers can use static routes only to route
>  to their leaf clients;
>
>- the backbone routers should not advertise default prefixes
>  (e.g. ::0/0 5f00::/16) between themselves.  It is ok
>  to advertise default prefixes to leaf clients;
>
>- each backbone router should maintain a RIPng tunnel with
>  two or more other backbone routers.
>
>I think these, I hope, not too restrictive rules will improve
>overall 6bone connectivity as well as make lives router
>administrators easier.
>
>Dimitry