23 Aug
2013
23 Aug
'13
10:56 a.m.
On Fri, Aug 23, 2013 at 05:41:32AM -0700, Eric Fort wrote:
Why is having rip routes sent from multiple places
such a BAD thing?
It's not, as long as they have the same information. We've actually
discussed having the portal send rip44 as well, since it would be working
from the same encap database as amprgw.
Is it
not a GOOD thing to have a choice on multiple redundant routes from one
place on the mesh to another?
It might be but rip44 doesn't have that capability; if you receive different
routes from different rip44 senders they'll just flap back and forth because
there can only be one route to a destination at a time.
It could be considered an error to think of the encap table entry
as a 'route'; it's really a list of endpoints, and there is only one
endpoint per subnet possible in its definition. Rip44 is simply a way of
transmitting the encap table, not setting routes to those endpoints. The
actual routing is done by the standard operation of Internet routing
protocols at the "outer" encapsulation level.
everyone in the mesh ought to be running a dynamic
routing protocol thus
increasing redundancy and reliability.
This is a good idea. I'm not up on the latest RFCs concerning routing in
tunneled (VPN, GRE, etc) networks. Do you have a recommendation?
- Brian