The heart of EIGRP is a mathematical rule that lets a router prove a backup path is loop-free without seeing the topology. Every entry in the topology table carries two numbers: the reported distance (RD) — the neighbour's own metric to the destination — and the feasible distance (FD) — the best total metric this router has ever had to that destination. The rule, called the feasibility condition, is simply: a neighbour is a feasible successor if its RD is strictly less than the FD.
The logic is elegant: if my neighbour is closer to the destination than my entire best path, then my neighbour cannot possibly be routing through me — a path through me would make its distance larger than mine. No topology map required; loop freedom falls out of arithmetic.
Successor and feasible successor
- Successor — the neighbour on the lowest-total-metric path; installed in the routing table as the next hop.
- Feasible successor (FS) — any other neighbour satisfying the feasibility condition; pre-validated as loop-free and held in the topology table, ready for instant promotion.
- Everything else — paths that exist but fail the condition; usable only after a fresh DUAL computation proves them safe.
R1# show ip eigrp topology 10.99.0.0/24EIGRP-IPv4 Topology Entry for AS(100)/ID(1.1.1.1) for 10.99.0.0/24 State is Passive, Query origin flag is 1, 1 Successor(s), FD is 131072 10.1.12.2 (GigabitEthernet0/0), from 10.1.12.2 Composite metric is (131072/130816), route is Internal 10.1.13.3 (GigabitEthernet0/1), from 10.1.13.3 Composite metric is (196608/130816), route is Internal
Read each metric pair as (computed distance / reported distance). The second path's RD, 130816, is less than the FD of 131072 — so it is a feasible successor, and losing the successor causes a sub-second, query-free failover.
Passive, active, and the diffusing computation
A route with a working successor is passive — the healthy state. If the successor is lost and a feasible successor exists, EIGRP promotes it and stays passive. If no FS exists, the route goes active: the router sets the query origin flag, multicasts queries to every neighbour, and must receive a reply from each before it can safely choose a new path. Each queried neighbour either answers from its own table or, if it also loses its only path, propagates the query further — this is the 'diffusing' in Diffusing Update Algorithm.
The failure mode is the stuck-in-active (SIA) route: if any neighbour in the query diffusion fails to reply within the active timer (3 minutes by default, with SIA-query probes at 90 seconds), the adjacency with the silent neighbour is reset — a small failure escalated into a bigger one. Large, flat EIGRP domains with long query reach are SIA factories, which is exactly why the next chapter's tools exist.
Bounding the query scope
Query range is the primary EIGRP design consideration. Two mechanisms stop queries at a boundary: summarisation — a neighbour that only knows a summary answers a query for a component prefix immediately with unreachable, ending diffusion — and stub routing, where a stub router declares it has no downstream paths, so neighbours never query it at all. Design an EIGRP network so no query can travel more than a couple of hops, and DUAL delivers what it promises: loop-free convergence measured in milliseconds.