Feasible distance

[gornication]

Hi!

  Feasible distance - the metric value for the lowest-metric path to reach a destination.(official cert guide)
 

  So, the metric value for the lowest-metric. It seems that only successor route has the lowest metric. But in the same book you can find "successor route FD" and "feasible successor FD". So the question.

Can FD be an attribute of a route that is not a successor route?

[Giuseppe Larosa]

Hello @gornication ,

the more correct definition of feasible distance is a record of the lowest metric since last time the route for this prefix has gone active.

Coming to your question: the feasibility condition compares the reported distance = the metric as seen by the EIGRP neighbor of the local node, with the feasible distance of the local node.

It is also called AD advertised distance in some books.

RD < FD  for the neighbor to be selected as a feasible successor.

The reasoning is the following: to avoid routing loops an EIGRP router can select a feasible successor only if that neighbor is more near to the destination prefix then the local node so that in case of faliure of the current successor if the local node selects the feasible successor as new successor there is no risk that the new device can point to the local node for the prefix that would cause a routing loop.

The feasibile condition together with an ordered stream of EIGRP updates sent and ackwnoledged by each EIGRP neighbor allows EIGRP DUAL to be loop free.

Hope to help

Giuseppe

[gornication]

Thank you, Giuseppe.

Unfortunately, I did not find the answer to my question in your reply.

[Giuseppe Larosa · Accepted solution]

tHello @gornication ,

>> Can FD be an attribute of a route that is not a successor route?

No, a feasibile successor route is a route that satisfies the FC feasibility condition that is :

RD < FD

RD = Reported distance as explained in my previous post.

We can say that in EIGRP each prefix in passive state has its own FD and this  is used to discriminate between feasible successors and not feasbile successors on the alternate routes if any.

Hope to help

Giuseppe

[gornication]

That is, no, FD is exclusively the successor route attribute, but like that you can call a similar feasible successor route metric. 
This is not correct, but is this use sometimes acceptable?

I understand that in the context of feasibility condition the feasible distance must be defined exactly how the feasibility condition of the successor route. But when, for example, we are talking about variance, other successor route FD and feasible successor FD are more convenient.

[Richard Burts]

It is my understanding that FD is not an attribute of the successor route. It is an attribute of the prefix for which we are looking for routes. Let me suggest an example which might help clarify the concept. Let us have routerA which is looking for a route for 192.168.10.0/24. routerA has 2 neighbors, routerB and routerC. routerA goes active for the prefix and sends queries for 192.168.10.0/24 and receives responses from routerB and from routerC. routerA evaluates the responses and the route from routerB is better than the one from routerC. So routerA selects the path through routerB and calculates the FD. routerC meets the feasibility condition and qualifies as a feasible successor.

Now there is some network event that impacts routerB and that path is no longer viable. routerA removes the route using routerB and since there is a feasible successor routerA immediately puts into the routing table the route through routerC. So now routerC is the successor. But the FD does not change. FD is not an attribute of routes from either routerB or routerC. It is an attribute of the prefix 192.168.10.1/24.

The only time that routerA changes the FD of the prefix is when routerA has to go active for that prefix.

[gornication]

Thanks for the answer, Rick. This is already interesting. 

  If from the loss of communication through router B feasible successor route does not become successor route, then the metric (let's say) of feasible successor route does not become FD. I have no contradictions here.
But I just don't understand that FD is a prefix attribute. FD is a metric. Metric is an attribute of the route towards the prefix.

  I looked into the RFC:
Being effectively a record of the smallest known metric since the last time the network entered the PASSIVE state, the FD is not necessarily a metric of the current best path. Exactly one FD is computed per destination network.

  I understand this so that as long as there is a feasible successor route, while there is no need for DUAL, the distribution of roles does not change. Well, it follows that the answer to my question is NO.) 

1. Usually, cases of inaccessibility of the subnet through the successor route are described. But how will the emergence of a more profitable route affect this scheme? I'm trying to figure out what FD can change.

2. "FD is not necessarily a metric of the current best path.", - means the point in time when traffic is on the route?

[Richard Burts]

I am not clear what you are asking here. You ask "But I just don't understand that FD is a prefix attribute. FD is a metric. Metric is an attribute of the route towards the prefix". Let me try to explain this way: FD is the historical record of the best metric to the destination at the last time EIGRP went active for that destination. It is a value to which other metrics can be compared. It is not an attribute of any specific route. Would it help if instead of saying that FD is an attribute of the destination that I said that FD was a characteristic of the destination? Part of my point is that FD is associated with a destination and not associated with any specific route to that destination. 

You had mentioned variance in EIGRP. The use of variance does not change anything about FD. Note that FD is used in the implementation of variance. With variance you specify a multiplier and any route whose metric falls into the range specified by variance can be considered for insertion into the routing table. But the route is not inserted into the routing table just because it falls into the range. First the candidate route is evaluated against the feasibility condition (which uses FD). Only if the candidate route satisfies the feasibility condition is the route inserted into the routing table. 

I am also not clear about this "But how will the emergence of a more profitable route affect this scheme? I'm trying to figure out what FD can change." I believe that you are asking what if a new route becomes available (is advertised from some neighbor) and the metric through the new route is better than the route that is currently in the routing table (successor route). In that case EIGRP will replace the existing route table entry with the new entry (with its new metric) and the FD remains the same. 

You also ask "FD is not necessarily a metric of the current best path.", - means the point in time when traffic is on the route?" It might be a point in time when traffic is on the route. It might also be a point in time when you manually check the content of the routing table. The real point here is that FD was established when EIGRP completed the active process and selected a route to the destination. It is saying that the chosen route might change over time, but that FD does not change just because EIGRP chose a new route to the destination. 

[gornication]

Sorry, but I can’t find any reason to think that the FD is bound to a subnet and not a route to that subnet. 

"Any feasible successor’s FD with a metric below the EIGRP variance value is installed into the RIB.", - this is a quote from the book. "feasible successor’s FD", - because of this phrase, I began to have doubts about whether I understood FD correctly.

I used to think it was the metric for the lowest-metric path to reach a destination. Now it turned out that the most profitable route is not associated with FD.

The unavailability of the FS does not change the FD. The new more profitable route does not change the FD. You understood me correctly. I don't understand what event changes the FD. Now it's hard for me to say what FD is.

[Giuseppe Larosa]

Hello @gornication ,

>> I don't understand what event changes the FD

When the prefix successor fails and there is no feasible successor EIGRP DUAL moves the prefix to the Active state and sends out Queries about the prefix to look for a new path to reach the prefix if any exists.

From all the Replies the local nodes will calculate the new FD in addition to selecting a new successor.

In this moment the FD is equal to the best metric via the current successor.

If later there is an event that changes the best metric to the prefix without the need to go Active again the FD does not change but it tracks the previous "best metric" to the prefix.

This is why the FD is an attribute of the prefix and not an attribute of the current best route to the prefix.

Most of the times the two match but not always.

To be honest until recent books there was a lot of confusion about FD. Last books and the RFC have cleared that FD for a prefix can change only during a passive to active to passive state transitions.

Hope to help

Giuseppe

[Richard Burts]

@Giuseppe Larosa responded first and is spot on. The only event that changes FD is EIGRP transition to active for the prefix. I appreciate his explanation that FD is an attribute of the prefix and not of a route to the prefix. I also appreciate his comment about recent publications being much better about FD and that some earlier publications were not as good (and some not correct) in their treatment of FD. Well done Giuseppe

[gornication]

Well, DUAL changes the FD. Thank you for clarifying.


FD could be the previous "best metric" of the previous best route to the prefix? )

Defined as the least-known total metric to a destination from the
      current router since the last transition from ACTIVE to PASSIVE
      state.

This is from the RFC. Least-known I understand as the smallest metric for the direction. Usually the best route has the lowest metric. But this is at a certain moment and such a time reference explains why the FD is not necessarily a metric of the current best path, which appeared on the router later.

When the new best route disappears (besides SR and FSR), how does the switch take place, does the traffic go through FS again? What if there are two of them and after one is unavailable there will be another one that is better than FS. How is the absence of loops checked? Does DUAL start?


Thanks you

[Richard Burts]

I find the wording of the section that you post a bit puzzling. I would agree that least-known would suggest the smallest metric. But that is not how it works. If least-known really meant the smallest metric known then if the current route (from which FD had been calculated) changed and the new route had an even lower metric then the least-known would suggest that FD would change. But it does not. The only thing that changes FD is the transition from active to passive.

There is one part of your post that is easy to answer. "How is the absence of loops checked? Does DUAL start?" The absence of loops is checked in the feasibility condition without starting DUAL. If a route passes the feasibility condition then its Reported Distance is less than the FD. And if its reported distance is less than the FD then its route can not be through this router and therefore can not produce a loop.

[gornication]

I imagined the following picture:

New arrival route metric
...
SR with the best metric immediately after transition to passive state. This metric gives rise to a number that is used for technical purposes of the EGRP - FD.
FSR with metric of the route satisfying the feasibility condition. 
...
DUAL S and FS are not available and you have to look for a new routing chance.

The metric of the route to the successor and the FD are still strongly associated, even equivalent. But for a more successful description of the life of an EIGRP-router, it is sometimes possible to FD into a separate entity associated with a prefix, and not with a metric of the successor route.

I did not fully understand everything that you wrote. But you helped me a lot, and Giuseppe has already given the answer to the main question.

Thank.

[Richard Burts]

In terms of the picture that you imagine you say this "DUAL S and FS are not available and you have to look for a new routing chance" In the case where the successor and the feasible successor are no longer available then EIGRP transitions that prefix to the active state and initiates queries to its neighbors looking for a new route. 

Then you say "The metric of the route to the successor and the FD are still strongly associated" I am not clear whether this is a continuation of your picture or whether this is a new thought. If it is a continuation of the picture than I would say that when the successor and the feasible successor are no longer available then the metric of the route is no longer associated and EIGRP looks for a new route and a new FD. 

And then you say " it is sometimes possible to FD into a separate entity associated with a prefix, and not with a metric of the successor route." and I do not understand what you mean. You mention FD as if it were an action. Did you mean it is sometimes possible to route into a separate entity? If not that then what? And what kind of separate entity did you have in mind? Perhaps a subnet which is part of a network for which there is a route? Or perhaps a host within a subnet for which there is a route. You certainly can route to those entities, but the route used would have the metric for the successor for the larger entity. Or perhaps I am on the wrong track entirely. In which case please clarify.