Poison Reverse

wat08

I see no reason why a router can't just send a triggered partial update of a failed route, remove the failed route from it's routing table, and start the hold-down timer immediately.

Why must the router wait to receive an advertisement of a route that it already knows has failed before starting the hold-down timer? It seems unnecessarily repetitive.

And another random question:

What the hell happens when you assign a mask that is < 24 bits to a Class C address as shown in this article under the "Prefix Lengths" heading? (192.168.32.0/19) Is that allowed?

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094823.shtml

tech-airman

wat08,

wat08 wrote:

I see no reason why a router can't just send a triggered partial update of a failed route, remove the failed route from it's routing table, and start the hold-down timer immediately.

Which routing protocol are you talking about?

wat08 wrote:

Why must the router wait to receive an advertisement of a route that it already knows has failed before starting the hold-down timer? It seems unnecessarily repetitive.

Which routing protocol are you talking about?

wat08 wrote:

And another random question:

What the hell happens when you assign a mask that is < 24 bits to a Class C address as shown in this article under the "Prefix Lengths" heading? (192.168.32.0/19) Is that allowed?

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094823.shtml

It depends if the routing protocol supports CIDR or not. Keep in mind that CIDR is NOT the same as "CIDR notation."

wat08

RIP.

tech-airman

wat08 wrote:

RIP.

wat08,

Which version?

wat08

tech-airman,

Pretty sure that is irrelevant since both RIPv1 and RIPv2 make use of poison reverse. In fact, this is a question geared towards distance-vector routing protocols in general.

Nevermind, it was a question sparked by unnecessary curiosity. I've already lost interest and will just accept that the algorithm for what it is.

dtlokee

I think your original question may be based on the idea that the failed route is directly attached to the router, but what if it is not?

Question 2: The network command will assume classful behavior so if you had an interface with an IP address of 192.168.3.254/22 then you would need to use "network 192.168.3.0" even though the "network address" would be 192.168.1.0/22

tech-airman

wat08 wrote:

tech-airman,

Pretty sure that is irrelevant since both RIPv1 and RIPv2 make use of poison reverse. In fact, this is a question geared towards distance-vector routing protocols in general.

Nevermind, it was a question sparked by unnecessary curiosity. I've already lost interest and will just accept that the algorithm for what it is.

wat08,

It is relevant because RIPv1 waits for periodic broadcast updates from the RIPv1 routers directly connected to them. If a network goes down under RIPv1 then it just waits until the next periodic broadcast update to notify other RIPv1 routers of that fact through poison reverse. However, for RIPv2 there are both periodic multicast updates as well as the "triggered update" feature that you mentioned. In the case of a RIPv2 triggered update, the specific route is "poison reversed" by advertising a metric of 16 for the network that just went down. However, even with the triggered update being used, the hold down timer is still useful because what if the network comes back up in two update cycles? In that case, RIPv2 would advertise a metric less than 16 in it's periodic updates that the network that was previously down is now up. This is why it is highly relevant which version of RIP you're talking about.

wat08

tech-airman wrote:

If a network goes down under RIPv1 then it just waits until the next periodic broadcast update to notify other RIPv1 routers of that fact through poison reverse.

RIPv1 makes use of both poison reverse and triggered updates:

To get triggered updates, we simply add a rule that whenever a gateway changes the metric for a route, it is required to send update messages almost immediately, even if it is not yet time for one of the regular update message.  (The timing details will differ from protocol to protocol.  Some distance vector protocols, including RIP, specify a small time delay, in order to avoid having triggered updates generate excessive network traffic.)

"split horizon with poisoned reverse", and "triggered updates"

RIPv1 RFC: http://www.rfc-editor.org/rfc/rfc1058.txt


I wasn't trying to debate the usefulness of the hold-down timer, rather, the time in which it is applied. It makes sense now. Thanks.