OSPF is the interoperable default, but on an all-Cisco network EIGRP is often the faster, simpler answer — a distance-vector protocol with link-state manners. This covers how it actually picks routes, the feasible-successor trick that gives sub-second failover, unequal-cost load balancing, and the honest note on why RouterOS can't join the party.
EIGRP is Cisco's advanced distance-vector protocol — it shares the simplicity of distance-vector (neighbours tell you their best routes) but adds link-state-like features: incremental updates, sub-second convergence, and no periodic full-table floods. It computes a composite metric from bandwidth and delay (by default), and its killer feature is the DUAL algorithm, which pre-computes a loop-free backup path — the feasible successor — so a link failure fails over without any recomputation at all.
Historically it was Cisco-proprietary, which is exactly why it doesn't appear in the translation guide: RouterOS (and everyone else) never implemented it. Cisco published it as informational RFC 7868, but in practice it's still Cisco-only — so EIGRP is the one routing choice that locks you in. Use it when the estate is all-Cisco and you want its convergence and easy load-balancing; use OSPF or BGP when interop matters.
01
Modern IOS prefers "named" EIGRP configuration — one hierarchical block instead of scattered classic commands. Advertise interfaces, and neighbours form automatically.
R1(config)# router eigrp CORP R1(config-router)# address-family ipv4 unicast autonomous-system 100 R1(config-router-af)# network 10.0.0.0 0.0.0.255 # which interfaces run EIGRP R1(config-router-af)# network 10.1.0.0 0.0.255.255 R1(config-router-af-topology)# exit-af-topology # Neighbours on those interfaces adjacency up automatically (same AS + K-values). R1# show ip eigrp neighbors
Two routers form an EIGRP neighbourship only if they share the same autonomous-system number and the same K-values (the metric weights). Mismatch either and the adjacency silently never forms — show ip eigrp neighbors is just empty, with no obvious error. Also passive-interface the LAN-facing ports you don't want to peer on (or an attacker on the LAN can inject routes) — the same discipline as OSPF. Don't touch the K-values unless you know exactly why; the defaults (bandwidth + delay) are almost always right.
02
EIGRP's fast convergence comes from DUAL pre-computing a guaranteed loop-free backup. Understanding feasible distance vs reported distance is understanding EIGRP.
For each destination EIGRP tracks two numbers: your feasible distance (FD) — your best metric to it — and each neighbour's reported distance (RD) — their metric to it. The best path's next-hop is the successor. A backup neighbour qualifies as a feasible successor only if its RD is less than your current FD (the "feasibility condition") — which mathematically guarantees it isn't routing back through you, so it's loop-free and can be installed instantly on failure.
R1# show ip eigrp topology P 10.2.0.0/16, 1 successors, FD is 3072 via 10.0.0.2 (3072/2816), GigabitEthernet0/0 # successor via 10.0.0.6 (5120/2816), GigabitEthernet0/1 # feasible successor (RD 2816 < FD 3072) # Failure of the successor -> the feasible successor is promoted with # ZERO recomputation. That's the sub-second convergence.
If there's no feasible successor when the successor dies, EIGRP can't just promote a backup — it sends queries to its neighbours asking for a path, and the route goes Active until they answer. Usually fast, but in a large flat network a query can propagate far and you get "Stuck In Active" (SIA) events. The fix is the same one that helps OSPF: summarisation and stub routers to bound how far queries travel. Design so common failures always have a feasible successor.
03
OSPF only load-balances across equal-cost paths. EIGRP can balance across unequal-cost paths — a genuinely useful trick for dual-WAN of different speeds — with one command.
# variance N means: also install any feasible successor whose metric is # within N times the successor's FD. Traffic is shared in proportion. R1(config-router-af-topology)# variance 2 # With variance 2, a path up to 2x the best metric also carries traffic, # roughly proportional to its cost. Only feasible successors qualify — # it can never load-balance onto a potentially-looping path. R1# show ip route eigrp # both next-hops now installed
04
Unlike OSPF, EIGRP can summarise on any interface, not just area borders — which bounds query scope and shrinks tables. Then confirm neighbours, topology, and the installed routes.
# Advertise one aggregate out an interface instead of many specifics: R1(config-router-af)# af-interface GigabitEthernet0/1 R1(config-router-af-interface)# summary-address 10.1.0.0 255.255.0.0 # --- Verify --- R1# show ip eigrp neighbors # adjacencies up, uptime climbing R1# show ip eigrp topology # successors + feasible successors R1# show ip protocols # AS, K-values, networks, passive ifaces R1# show ip route eigrp # D routes in the table
Takeaways
variance gives unequal-cost load balancing — use two different-speed links proportionally, something OSPF can't do.NOCTIS designs and tunes Cisco routing — EIGRP, OSPF, and BGP — across Crete, with fast failover, bounded query scope, and load-balancing that actually uses every link you pay for.
Book a Discovery Call →