EIGRP question

Unanswered Question
Jun 14th, 2009

Is it possible that an EIGRP router selects a sub-optimal route over an optimal route?

|-----5---R2----5-------|

R1-----2---R3----12---R5----|Net-A

|-----9---R4----6-------|

The cost is simplified for ease of calculation.

- Based on the above scenario, the best route to Network-A from R1 is through R2; therefore the Feasible Distance equals 11 (5 + 5 + 1 = 11).

- Only route from R1 to Network-A through R4 satisfies the Feasible Condition, and is installed in the Topology table.

- Route R1-R3-R5-Network-A is left out b/c it does not meet the FC, even though it has better metric to reach Network-A than through R4.

Now consider how R5 routes packet back to R1. The best path back to R1 is through R2 with FD = 10.

R5 also have 2 other routes back to R1 through R3 and R4, both of which satisfy the FC.

Let's say R2 fails, R1 would look into its topology table and select a backup route through R4 (the only backup route). R5 would do the same thing and select backup route through R3 (better metric). Would the result of this be an assymetric routing path from R1 - Network-A and back??? R1 - R4 - R5 - R3 - R1

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Giuseppe Larosa Sun, 06/14/2009 - 22:19

Hello Thong,

the feasible condition is meant to avoid routing loops also during transitions: the selected candidate next-hop has to be nearer to the destination then the local node, so that it (the potential next-hop) will never try to use the local node as its next-hop for the destination.

Asymmetric route paths are not a problem until stateful firewall devices are not on the path (if a device wants to see both directions of a flow to decide if it is legitimate).

Hope to help

Giuseppe

thodao Sun, 06/14/2009 - 22:59

Hi Giuseppe, thanks for replying to my post. My question being if EIGRP in some cases would choose a sub-optimal path over the optimal one (from metric perspective) because the sub-optimal path satisfies the FC? If you configure my scenario using OSPF, I think it would use the path R1-R3-R5 if R2 failed instead of R1-R4-R5 like it does in EIGRP. There is a possibile flaw in my logic in the orginal post though.

Thanks again :-)

Giuseppe Larosa Sun, 06/14/2009 - 23:39

Hello Thong,

the scenario is interesting even if in real world using both delay and bandwidth it is difficult to get something similar: metrics are usually very high values and slightly different.

Be aware that EIGRP metric is cumulative in delay and inverse proportional to lowest BW on path.

As I wrote in my first post the FC is meant to select only candidate next-hops that are nearer to the destination then the local node that in other words they will never use local node to reach the subnet.

I agree that OSPF logic is different.

Hope to help

Giuseppe

Actions

This Discussion