Route Redistribution Question
Hello All,
I'm new here; been browsing and reading up as a guest for awhile but now I hit a wall and cannot find the answer in plain text!
When redistributing routes I understand that there are two steps:
1) Routes "learned" in routing table are redistributed
2) Interfaces participating in the Routing protocol are redistributed
If you use a policy map to define a specific route to be redistributed... oh lets say the loopback interface they always want you to configure; it changes the default behavior in the sense that #2 no longer happens and now your redistributions do not include any connected interfaces that participate in the Routing Protocol (but are not learned from the routing protocol).
My question is; if instead of redistributing a policy-map, you redistribute static routes. Will the same behavior be observed?
Also,
Is there a reason why when redistributing into EIGRP areas that the "subnets" command is not needed?
Thanks guys!
I'm new here; been browsing and reading up as a guest for awhile but now I hit a wall and cannot find the answer in plain text!
When redistributing routes I understand that there are two steps:
1) Routes "learned" in routing table are redistributed
2) Interfaces participating in the Routing protocol are redistributed
If you use a policy map to define a specific route to be redistributed... oh lets say the loopback interface they always want you to configure; it changes the default behavior in the sense that #2 no longer happens and now your redistributions do not include any connected interfaces that participate in the Routing Protocol (but are not learned from the routing protocol).
My question is; if instead of redistributing a policy-map, you redistribute static routes. Will the same behavior be observed?
Also,
Is there a reason why when redistributing into EIGRP areas that the "subnets" command is not needed?
Thanks guys!
Comments
-
Forsaken_GA
Member Posts: 4,024
Hello All,
If you use a policy map to define a specific route to be redistributed... oh lets say the loopback interface they always want you to configure; it changes the default behavior in the sense that #2 no longer happens and now your redistributions do not include any connected interfaces that participate in the Routing Protocol (but are not learned from the routing protocol).
My question is; if instead of redistributing a policy-map, you redistribute static routes. Will the same behavior be observed?
You mean route-map instead of policy map I assume.
Connected interfaces are funny because they can be categorized as either connected, or actual routing protocols routes.
What happens is that when you use a route-map to match which interfaces you want to redistribute, there's a catchall at the end that denies anything you don't allow, much like an access-list.
Since a protocol is not going to distribute a route twice, it has an order of precedence - an interface that's participating in the routing protocol is considered a connected interface before it's considered a routing interface.
So if I use a route-map to match and redistribute lo0, and don't define anything else, it's the same as saying permit lo0 deny all. If I then later redistribute a routing protocol that has interfaces which have been previously denied due to connected redistribution, I don't get those routes unless I start modifying route-maps to allow them.
Static routes, in contrast, are a bit easier. There's no dual nature with a static route. redistribute static will redistribute any valid static routes, unless you run it through a route-map to filter what you want to allow redistributed.Also,
Is there a reason why when redistributing into EIGRP areas that the "subnets" command is not needed?
Because that's the way EIGRP was designed. OSPF only redistributes major classful networks by default because that' the way it was designed. EIGRP was not, it redistributes what is actually in the routing table.