skip to content

Why does an EIGRP router keep a topology table as well as its routing table, and what does each one hold?

level: juniorimportance: should knowfreq 32%

answer

  1. every neighbour's advertisement, kept
  2. reported versus computed distance
  3. only successors are installed
  4. backup ready before the failure

basics

~10 s

The EIGRP topology table records every path any neighbour advertised for each destination, with its distances and route state; the routing table receives only the successor paths actually used to forward packets.

solid answer

~40 s

EIGRP keeps two different things. The **topology table** stores, for every known destination, each neighbour that advertised it, that neighbour's **reported distance** (RD), the router's **computed distance** (CD) through it, the destination's **feasible distance** (FD) and whether the route is passive or active. The **routing table** gets only the **successors**, the least-cost loop-free next hops. Keeping the losing paths is the point: a backup that already passes the loop-freedom test (a **feasible successor**) can be promoted the moment the successor fails, with no need to ask the neighbours first. Paths that fail that test stay in the topology table too, but they are never installed while the route is passive.

go deeper

for a junior

Recall that EIGRP keeps every neighbour's advertisement in its topology table and installs only the best loop-free path, the successor, in the routing table.

for a middle

Explain an entry: each neighbour's reported and computed distance, the destination's feasible distance and route state, and how successors and feasible successors are picked from them.

for a senior

Tie the second table to convergence: a feasible successor is promoted locally without queries, and show why paths that fail the loop-freedom test are kept but never forwarded on.

for a principal

Frame the design choice: EIGRP spends memory on per-neighbour state to buy local failover, where a link-state protocol keeps a full map and recomputes; weigh that against the operational cost of a vendor-originated protocol.

## Two tables, two jobs EIGRP is a distance-vector routing protocol, specified publicly as Informational RFC 7868 after starting life as one vendor's protocol. Like every distance-vector protocol, it learns routes from what its neighbours tell it rather than from a map of the whole network. What sets it apart is that it does **not** throw away the advertisements it decides not to use. It keeps two separate structures: - The **topology table** is EIGRP's own working store. RFC 7868 defines it as the structure holding every known destination with its prefix and prefix length, the **feasible distance** (FD), the **reported distance** (RD) of each neighbour advertising it, the **computed distance** (CD) over that neighbour, and the **route state**. - The **routing table** is the router's forwarding decision for every protocol together. EIGRP contributes only the paths it has chosen to forward on. ## The vocabulary of one entry | Term | Meaning | How many | |---|---|---| | Reported distance (RD) | The neighbour's own distance to the destination, as it advertised it | one per neighbour | | Computed distance (CD) | The total distance from this router through that neighbour, built from its RD and the cost of the link to it | one per neighbour | | Feasible distance (FD) | The lowest distance this router has known for the destination since the route last became passive | one per destination | | Successor | A neighbour that provides the least-cost path and passes the loop-freedom test | one or more | | Feasible successor | A neighbour that passes the loop-freedom test, whatever its cost | zero or more | Tools that display the table often show each path as a pair, its total and its reported distance; the total is the CD in the RFC's words. ## What reaches the routing table 1. EIGRP collects every neighbour's advertisement for the destination into the topology table. 2. It computes a CD through each neighbour. 3. The neighbour (or neighbours, on a tie) offering the lowest CD that also passes the **feasibility condition** becomes the **successor**. RFC 7868 says DUAL, the Diffusing Update Algorithm, selects the routes inserted into the routing table based on feasible successors, and calls the least-cost ones successors. 4. Only successors are offered to the routing table. Where several protocols offer the same prefix, the router still chooses between sources by its own ranking (administrative distance), a separate step. The feasibility condition itself, a neighbour's RD strictly below this router's FD, belongs to DUAL; the topology table is where the numbers it compares live. ## Why keep the paths it is not using The payoff is convergence. If the successor's link fails and the topology table already holds a **feasible successor**, the router promotes it to the routing table on the spot. The feasibility condition has already guaranteed that path is loop-free, so no coordination with other routers is needed and the switchover is local. When no feasible successor exists, the route goes active and DUAL starts asking neighbours, which is slower and is DUAL's business rather than the table's. Paths that fail the feasibility condition are also kept. RFC 7868 notes that routes from neighbours that may be upstream are not recorded in the routing table but are saved in the topology table: the router knows they exist, but cannot prove they do not loop back through itself, so it does not forward on them while the route is passive. ## Where candidates go wrong - Treating the topology table as a link-state map of the network. It is not: it holds only what directly connected neighbours reported, per destination. A full map is a link-state protocol's database. - Assuming the feasible successor is pre-installed in the routing table as a hot standby. It is not installed until it is promoted, unless an implementation's unequal-cost load sharing is turned on and admits it. - Believing the topology table holds only feasible paths. It holds every advertisement; the routing table is the filtered view. - Confusing FD with "the current best metric". In a stable network they are equal, but FD is a historical record that can be lower than the current best after a path gets worse. ## Why interviewers ask The question checks whether a candidate understands EIGRP's fast-convergence claim as a consequence of data structure: the backup is computed and proven loop-free before it is needed. A good answer names both tables, states what each entry carries, and says that only successors are installed.

  • In EIGRP, is a feasible successor installed in the routing table alongside the successor?
    Not normally. While the route is passive and only equal-cost forwarding is in use, the routing table holds the successor or successors; the feasible successor waits in the topology table. It is installed when it is promoted after the successor fails, or earlier only if an implementation's unequal-cost load sharing is enabled and admits it.
  • Why does an EIGRP router keep paths in its topology table that fail the feasibility condition?
    They are still real advertisements: the neighbour may be closer to the destination after the topology changes, and DUAL needs every neighbour's current distance when it recomputes. The router simply cannot prove such a path is loop-free right now, so it records it but does not forward on it while the route is passive.

Like keeping every contractor's quote on file with the chosen one pinned on the noticeboard: a runner-up already vetted can be pinned the moment the first drops out, without calling round for new quotes.

saying these in an interview costs you the question

  • The EIGRP topology table is a full link-state map of every router and link.
  • The feasible successor is already installed in the routing table as a standby.
  • The topology table only stores paths that passed the feasibility condition.
  • Feasible distance always equals the current best metric to the destination.
  • Every EIGRP path in the topology table is used for forwarding.