EIGRP Split horizon over frame Relay + difference between successor and Feasible distance

[Steph1963]

Hi,

I  am reviewing the following  document

http://www.cisco.com/en/US/tech/tk365/technologies_white_paper09186a0080094cb7.shtml

and I have some dificulties to understand the two following items:

Par 1:How does EIGRP use the concepts of feasible distance, reported distance, and feasible successor to determine if a path is valid, and not a loop? In Figure 4a, Router Three examines routes to Network A. Since split horizon is disabled (for example, if these are multipoint Frame Relay interfaces), Router Three shows three routes to Network A: through Router Four, through Router Two (path is two, one, three, four), and through Router One (path is one, two, three, four).

Question: I thought split horizon was enabled by default over frame-relay. Does this means that each interface would be configured as different multipoing sub-interface. Why would we want to do this, does using logical point-to-point sub-interface would do the same since there is only one conncection per sub-interface.

Par 2: For instance, in Figure 4a, if Router One is connected to Routers Two and Three through a single multipoint interface (such as Frame Relay), and

Question 2: Does this refer to the default behaviour of frame-relay of enabling split horizon on multipoint interface.

Question 3: What is the difference between successor and feasible distance.

Thanks for your help

Stephane

[Peter Paluch]

Hi Stephane,

Question: I thought split horizon was enabled by default over
frame-relay. Does this means that each interface would be configured as
different multipoing sub-interface. Why would we want to do this

First of all, the entire situation description of the Figure 4a in the article you referenced is artificial. It is an improbable scenario you would probably never see in a real network. It is constructed by simply forgetting about some additional mechanisms (like split horizon and/or poisoned reverse here) to concentrate simply on the EIGRP logic itself.

Just to complete the discussion about split-horizon: for EIGRP, the split horizon is active on a Frame Relay multipoint interface by default. However, in a hub-and-spoke Frame Relay technology, the split horizon is not a welcome technique because it will prevent  a network learned from one spoke to be advertised by the hub to other spoke routers. Therefore, in such networks, the split-horizon must be explicitly deactivated.

Of course, if we modified the configuration to convert the multipoint interface on the hub router to a number of point-to-point subinterfaces (one for each PVC) then we would not need to modify the split horizon settings because it would not apply. The downside of this solution is the actual need to create so many interfaces and possibly spend quite a lot of IP addresses on distinct networks on these interfaces.

Question 2: Does this refer to the default behaviour of frame-relay of enabling split horizon on multipoint interface.

Yes, it does, as I described earlier. Split horizon does not allow you to advertise the network out the interface through which you learned if that interface happens to be the one you are using to reach that network (i.e. the interface is on the best path towards the destination).

Question 3: What is the difference between successor and feasible distance.

A successor and feasible successor are your EIGRP neighbors (directly connected routers) that fulfill certain criteria. The feasible distance is a metric towards a particular destination also meeting special criteria.

A successor must meet two conditions:

1. It must not be on a routing loop
2. It must provide the lowest total distance to the destination

A feasible successor must meet a single conditions:

1. It must not be on a routing loop

Now, if you look at it carefully, both successor and feasible successor can be used to reach a network. However, the total path through the feasible successor is not the best available.

A feasible distance is the minimum distance to the destination since the last time the destination went from the active to the passive state. It is thus a record of the length of the shortest path to the destination since the last time we have found a successor for that route. In passive state, it can only decrease (by its definition as the minimum of all distances seen to the destination since the last active->passive transition). If it is necessary to increase the feasible distance, the router must go into the active state and commence a diffusing computation.

Best regards,

Peter

[Steph1963]

Hi Peter,

Took me a while to figure it (I hope), can you just please confirm me that my understanding is fine.

The successor is the path that will be used by the routing table. Router that receives information from the successor just add the interface delay that receive the announcement to update the metric (or modifiy it if bandwidth is smaller).

The following show ip eigrp topology is showing two feasible successor (only meaning that they are not part of any loop) and the best path is viaS0/0

P 172.16.16.0/24, 1 successor, FD is 284244

via 172.16.250.2 (284244/17669856), Serial 0/0

via 172.16.250.1 (12738176/27819002), Serial 0/1

The feasible successor only indicates that more than one route are available to a specific path but does not guarantee it since it is possible that the alternate route has a high cost to that route.

Thanks for your help

Stephane

[Peter Paluch]

Hello Stephane,

No problem - the EIGRP internals are quite obscure.

The successor is the path that will be used by the routing table.

Essentially, yes. To be terminologically precise: The path through a successor will be used by the routing table. As I mentioned earlier, the successor is a neighbor, not a path.

Router that receives information from the successor just add the
interface delay that receive the announcement to update the metric (or
modifiy it if bandwidth is smaller).

Yes - the metric as received (reported) by the current successor will be modified using the interface metric that received this update, i.e.

• The bandwidth will be lowered if the interface has a smaller bandwidth than the bandwidth in the EIGRP update
  
• The delay of the interface will be added to the delay in the EIGRP update

The following show ip eigrp topology is showing two feasible
successor (only meaning that they are not part of any loop) and the
best path is viaS0/0
P 172.16.16.0/24, 1 successor, FD is 284244
via 172.16.250.2 (284244/17669856), Serial 0/0
via 172.16.250.1 (12738176/27819002), Serial 0/1

Well, this output puzzles me. According to my experience, the show ip eigrp topology output always contains information in the format of:

via NEXT-HOP (Total Distance/Reported Distance)

that implies that the number before the "/" sign shall be strictly higher than the number after it. Yet, your output does not conform to this. Is that by accident or does your router really display it this way? I cannot really comment here what is what because I do not understand this output, i.e. what is the Total and the Reported Distance here.

The feasible successor only indicates that more than
one route are available to a specific path but does not guarantee it
since it is possible that the alternate route has a high cost to that
route.

Essentially, yes - but I would not use the word "guarantee" here. A feasible successor already provides a perfectly valid, usable and workable path that is guaranteed to deliver your packets without any routing loops to their correct destination, only the total distance through a feasible successor is higher than the total distance through the current successor, so the path is not the shortest available.

Best regards,

Peter