Sign in Subscribe

Comparing EIGRP and OSPF for CCNP 350-401 ENCOR: How They Really Choose Paths, Converge, and Break in Production

Comparing EIGRP and OSPF for CCNP 350-401 ENCOR: How They Really Choose Paths, Converge, and Break in Production

1. Introduction: Why ENCOR Compares EIGRP and OSPF

ENCOR really doesn’t care much for surface-level memorization. Cisco wants you to understand what routing actually does when things break, when networks reconverge, when redistribution gets involved, and when the design itself is the problem. That is why EIGRP and OSPF appear so often in scenario questions. You are not just identifying an advanced distance vector protocol versus a link-state protocol; you are reading operational behavior from CLI evidence and predicting what the control plane will do next.

The high-value comparison is this: EIGRP uses neighbor relationships, a topology table, and the DUAL algorithm to maintain loop-free paths and possible backups. OSPF builds a per-area link-state database, synchronizes LSAs, and runs SPF to calculate shortest paths. And that architectural difference is exactly why the exam keeps coming back to path selection, load balancing, convergence, route types, administrative distance, summarization, redistribution, and troubleshooting.

EIGRP is not classic distance vector like RIP. It is commonly described as an advanced distance vector protocol. It does not flood a full topology database like OSPF, but it does maintain rich topology awareness through DUAL. It tracks successors, evaluates loop-free alternatives, and sends partial, bounded updates instead of full periodic table dumps.

OSPF is a true link-state protocol. Routers bring up adjacencies, exchange LSAs, keep LSDBs for their areas in sync, and then run SPF locally on each box. Be precise here: routers in the same area should have matching LSDBs for that area. ABRs maintain separate LSDB views for multiple areas, so there is no single global LSDB for the entire autonomous system.

Key comparison points include protocol model, primary algorithm, knowledge source, update style, and scaling method:

  • EIGRP: advanced distance vector, uses DUAL, learns through neighbors and the topology table, sends partial bounded updates, and scales through summarization, stub design, and query containment.
  • OSPF: link-state, uses SPF or Dijkstra, learns through a per-area LSDB, floods LSAs, and scales through area hierarchy, LSA scoping, and a stable backbone.

And honestly, interoperability matters a lot too. OSPF is broadly multivendor and remains the default choice in mixed environments. EIGRP was published as an informational standard, so it is no longer purely proprietary in the historical sense, but in practice it is still mostly a Cisco-centric design choice.

3. How They Build Routing Knowledge

EIGRP relies on three core structures: neighbor table, topology table, and routing table. The topology table is the important one for ENCOR because it stores feasible distance, reported distance, successor, and possible loop-free alternatives. A route is passive when stable and active when DUAL is searching for a new path.

OSPF uses a neighbor table, per-area LSDBs, the SPF result, and finally the routing table. OSPF does not simply trust a neighbor’s metric to a prefix. It learns the topology, then computes the shortest path itself.

That difference drives troubleshooting. If an EIGRP route is missing, I usually start thinking about no neighbor, no feasible successor, or a route that’s gone active and still waiting on DUAL. With OSPF, a missing route usually points me toward a bad adjacency, missing LSAs, an area issue, or just the route-type preference rules inside OSPF doing exactly what they’re supposed to do.

4. OSPF Neighbor States and DR/BDR Behavior

ENCOR loves OSPF adjacency traps, especially on Ethernet. The normal neighbor states are Down, Init, 2-Way, ExStart, Exchange, Loading, Full.

  • Down: no hellos received.
  • Init: hello received, but bidirectional communication not yet confirmed.
  • 2-Way: bidirectional hello communication established.
  • ExStart: master/slave and DBD negotiation begins.
  • Exchange: DBD packets exchanged.
  • Loading: LSRs and LSUs fill in missing database details.
  • Full: LSDB synchronization complete.

The exam trap: on broadcast and NBMA multiaccess networks, DROTHER routers normally remain in 2-WAY with each other. That is not a fault. FULL adjacency is normally expected with the DR and BDR, or on point-to-point links. Do not treat every 2-WAY state as a problem.

Common OSPF adjacency blockers include area mismatch, authentication mismatch, hello/dead timer mismatch, MTU mismatch, subnet mask mismatch on some network types, duplicate router IDs, and network type mismatch. MTU mismatch commonly leaves neighbors stuck in ExStart or Exchange.

5. EIGRP DUAL Deep Dive

DUAL is where EIGRP earns its reputation. Key terms:

  • Successor: current best path installed in the RIB.
  • Feasible successor: loop-free backup candidate.
  • Feasible distance (FD): local best metric to the destination.
  • Reported distance (RD), also called advertised distance: neighbor’s metric to that destination.

The feasibility condition is simple but critical: a candidate path is considered loop-free if the neighbor’s RD is less than the local router’s current FD. In shorthand: RD < current FD.

If the successor fails and a feasible successor exists, EIGRP can switch quickly. If no loop-free backup exists, the route goes active and the router sends queries to neighbors. Those neighbors reply if they know a valid path or must query further if they do not. Poor design lets this query scope spread too widely.

This is also where stuck-in-active (SIA) matters. If those replies don’t show up fast enough, EIGRP can mark the route stuck-in-active and, if things keep dragging, it may even reset the adjacency. That’s usually not some mysterious protocol failure. More often than not, it means the query scope got too big, the summarization isn’t tight enough, the links are slow, or the router’s just getting hammered.

EIGRP packet types are worth knowing: Hello, Update, Query, Reply, and ACK. RTP, or Reliable Transport Protocol, is what makes sure certain EIGRP packets get delivered reliably. For troubleshooting, that helps explain why adjacency, loss, and delay issues can affect control-plane behavior differently than simple data-plane tests suggest.

6. Metrics and Path Selection

EIGRP metric is composite. In classic operation, the default K-values are K1=1, K3=1, K2=K4=K5=0, so the metric is based on minimum bandwidth along the path and cumulative delay. Reliability and load exist in the formula but are not used by default. ENCOR usually tests the operational idea, not long manual arithmetic, but you should know the components.

Important units:

  • bandwidth on Cisco interfaces is in kbps.
  • delay is in tens of microseconds.

If K-values do not match between neighbors, adjacency will not form. That is a classic exam and troubleshooting point.

OSPF metric is cost. In Cisco implementations, cost is typically derived as reference bandwidth / interface bandwidth, using Mbps. Manual interface cost overrides the calculated value. If reference bandwidth is left at default in a modern network, multiple high-speed links can all collapse to cost 1, which breaks path differentiation.

Example OSPF tuning:

router ospf 10 auto-cost reference-bandwidth 100000

That value is in Mbps, so 100000 represents 100 Gbps. And it’s really important that you set it consistently across every OSPF router in the domain.

OSPF path choice is not just “lowest cost.” Within OSPF, route preference is broadly:

  • Intra-area preferred over
  • Inter-area preferred over
  • External

Among externals, E1 includes internal cost to the ASBR plus external cost, while E2 uses primarily the external metric and is the default in many redistribution cases. That distinction matters in exam questions.

7. Load Balancing: ECMP vs. Unequal-Cost Sharing

Native OSPF route calculation supports equal-cost multipath only. If paths do not have equal cost, OSPF does not load share across them as part of normal SPF behavior.

EIGRP supports equal-cost and unequal-cost load balancing with variance, but the path must still be loop-free. In exam language, think of variance as allowing paths that satisfy the feasibility condition and fall within the variance threshold. Also remember that maximum-paths can limit how many routes are installed.

In production, forwarding behavior is usually handled by CEF per-flow or per-destination hashing, not simplistic packet-by-packet round robin. That matters because unequal-cost sharing can create asymmetric application behavior even when the routing table looks clean.

8. Convergence, Summarization, and Scaling

EIGRP converges fastest when a feasible successor already exists. If not, DUAL must query. That is why stub routers and summarization matter so much. EIGRP stub routers limit what they advertise and, more importantly, help prevent them from being used as transit points for wide query propagation. Common stub options include connected, summary, redistributed, receive-only, and static.

Summarization also creates query boundaries. A summary router can answer on behalf of a range rather than letting every more-specific search continue across the domain. This is one reason EIGRP can scale very well in branch and hub-and-spoke designs when built intentionally.

OSPF convergence is driven by LSA flooding, LSDB update, and SPF recalculation. Bad area design and flaky links usually just create more churn and eat more CPU. Reference bandwidth inconsistency does not create more LSAs; it affects path choice only. Keep those issues separate.

Area design is the OSPF scaling tool. Area 0 is the backbone. ABRs connect areas and contain topology scope. Route summarization is typically done on ABRs for inter-area routes and on ASBRs for redistributed external routes. Stub, totally stubby, and NSSA designs reduce external LSA exposure and can shrink LSDB complexity. In NSSA, external routes appear as Type 7 LSAs inside the NSSA and are translated by the ABR to Type 5 for the rest of the domain.

Operational tuning exists too. Modern OSPF implementations may use SPF throttling, pacing, and in some cases incremental SPF behavior depending on platform and software, but the exam-safe model remains LSA change, LSDB update, SPF, then RIB update.

9. Route Types, LSAs, Administrative Distance, and RIB Decisions

Protocol best path is not always the route installed in the routing table. The RIB compares sources using administrative distance first.

  • EIGRP internal AD: 90
  • OSPF AD: 110
  • EIGRP external AD: 170

So in practice, an internal EIGRP route usually beats an OSPF route to the same prefix, while OSPF usually wins over an EIGRP external route.

For example, if 10.10.10.0/24 shows up as an internal EIGRP route with metric X and also as an OSPF route with cost Y, the router picks the EIGRP path because AD 90 beats 110. If that EIGRP route is external, though, OSPF wins because 110 beats 170.

When you’re troubleshooting OSPF, it really helps to keep the basic LSA types straight:

  • Type 1: Router LSA
  • Type 2: Network LSA
  • Type 3: Summary LSA from ABRs
  • Type 5: External LSA
  • Type 7: NSSA external LSA

That is enough for most ENCOR questions linking route types, area behavior, and LSDB content.

10. Configuration Model Notes and Practical Examples

Modern IOS XE often uses named EIGRP. If you’re working in named EIGRP mode, things like variance and passive-interface sit under the address-family submode. For example, it’d look something like this:

router eigrp CORP address-family ipv4 unicast autonomous-system 100 af-interface default passive-interface exit-af-interface af-interface GigabitEthernet1/0/0 no passive-interface exit-af-interface topology base variance 2 exit-af-topology

Classic mode is still common in study material, but do not mix classic numeric syntax with named-mode commands carelessly.

OSPF interface cost override example:

interface GigabitEthernet1/0/1 ip ospf cost 50

Here’s a simple EIGRP summarization example that makes the behavior easier to see:

interface GigabitEthernet1/0/0 ip summary-address eigrp 100 10.20.0.0 255.255.252.0

Remember that summarization may create a local summary route to Null0 to prevent loops if no more-specific route matches.

11. Redistribution Between EIGRP and OSPF

Redistribution is where clean theory becomes messy operations. EIGRP requires a seed metric for redistributed routes. OSPF redistribution should explicitly define metric and often metric type.

route-map OSPF-TO-EIGRP permit 10 set metric 100000 100 255 1 1500 set tag 110 router eigrp 100 redistribute ospf 10 route-map OSPF-TO-EIGRP route-map EIGRP-TO-OSPF permit 10 match tag 0 set metric 20 set metric-type type-1 set tag 100 router ospf 10 redistribute eigrp 100 subnets route-map EIGRP-TO-OSPF

The key idea is not memorizing one exact route-map. It is understanding the workflow: assign a seed metric, tag routes, and prevent feedback. Route tags aren’t magic by themselves; your policy still has to match them and actually do something useful with them. Whenever you can, keep redistribution narrow, explicit, and temporary. That’s just good operational hygiene.

12. Security and Stability Controls

Both protocols should be hardened. Passive interfaces stop hello exchange and prevent unintended adjacencies, though the connected network may still be advertised depending on protocol and configuration context. Use passive-interface by default and selectively enable only routing links.

EIGRP does support authentication, usually with key chains, and on newer platforms you might also see stronger options like HMAC-SHA-256. OSPFv2 often uses plaintext or MD5 authentication, depending on the platform and software version, and some implementations also support key-chain-based methods. The exam point is simple: authentication mismatch breaks adjacency, and no authentication leaves you exposed to unauthorized adjacency formation and route injection.

13. Troubleshooting Playbook

Use a structured workflow: neighbor formation → topology/database validation → route installation.

EIGRP commands:

  • show ip eigrp neighbors — verify adjacency; think AS mismatch, K-value mismatch, authentication, passive interface, subnet mismatch, ACLs.
  • show ip eigrp topology — verify successor, feasible successor, active state.
  • show ip protocols — verify process, networks, passive interfaces, redistribution.

OSPF commands:

  • show ip ospf neighbor — verify state and DR/BDR interpretation.
  • show ip ospf database — verify LSAs are present.
  • show ip ospf interface or show ip ospf interface brief — verify participation and interface parameters; note output support can vary by platform.

RIB check:

  • show ip route or protocol-specific views such as show ip route ospf

Useful debug tools, used carefully in labs or maintenance windows, include debug eigrp packets and debug ip ospf adj.

Two common scenarios:

  • EIGRP route goes active: likely no feasible successor, poor query containment, missing stub configuration, or a topology change that removed the successor. Look for active routes and query behavior.
  • OSPF neighbors stuck in ExStart/Exchange: often MTU mismatch. If two Ethernet neighbors are only 2-WAY and both are DROTHERs, that may be perfectly normal.

14. Design Guidance by Environment

EIGRP fits well in Cisco-heavy branch/WAN environments where stub routing, summarization, and flexible path control are valuable. OSPF fits well in large hierarchical or multivendor environments where area design and standards-based interoperability matter more than unequal-cost load sharing.

Neither protocol is “better” in the abstract. EIGRP problems usually come from uncontrolled query scope, weak summarization, or redistribution mistakes. OSPF problems usually come from bad area design, unstable links, adjacency mismatches, or poor cost tuning.

15. ENCOR Exam Traps and Final Review

  • EIGRP is advanced distance vector, not link-state.
  • OSPF is link-state and uses LSDBs and SPF.
  • EIGRP unequal-cost load balancing uses variance with loop-free eligible paths.
  • OSPF supports ECMP only in native route calculation.
  • On broadcast/NBMA networks, 2-WAY between DROTHER routers is normal.
  • EIGRP feasible condition: RD < current FD.
  • Default ADs to memorize: EIGRP internal 90, OSPF 110, EIGRP external 170.
  • Within OSPF: intra-area > inter-area > external; know E1 vs E2.
  • EIGRP scales with stubs and summarization; OSPF scales with areas and LSA containment.
  • Redistribution needs seed metrics, tags, and loop prevention.

If a question mentions DUAL, feasible successor, query, active state, variance, think EIGRP. If it mentions LSDB, SPF, DR/BDR, LSA type, area hierarchy, think OSPF. If the path in the routing table does not match the path you expected inside the protocol, think administrative distance and RIB installation logic first.