OSPF and IS-IS solve the same basic problem: distributing internal routing information using a link-state database and shortest-path calculations.[1][2] The useful interview question is not which acronym is better, but how their area boundaries, neighbor relationships and database exchanges differ.
Short answer: OSPF uses areas with a backbone called Area 0, while IS-IS uses Level 1 for routing inside an area and a connected Level 2 topology for backbone routing.[1][4] An OSPF ABR connects areas through its interfaces; an IS-IS L1/L2 router participates in both routing levels while belonging to its local area.[1][3]
OSPF vs IS-IS: key differences at a glance
Scope: the OSPF packet names and LSA terminology below refer to OSPFv2, and the IS-IS discussion concerns Integrated IS-IS for IP routing.
| Question | OSPF | IS-IS |
|---|---|---|
| What algorithm builds paths? | Link-state SPF calculation.[1] | Link-state SPF calculation.[2] |
| What defines hierarchy? | Areas connected through the Area 0 backbone.[1] | Level 1 areas interconnected by a contiguous Level 2 subdomain.[4] |
| Where is the area boundary? | At an ABR, whose interfaces attach to different areas.[1] | Between routers; an L1/L2 router belongs to its local area.[3] |
| Can neighbors have different area addresses? | Ordinary adjacency requires matching Area IDs on that link.[1] | L1 requires a shared area address; L2 does not.[4] |
| What identifies a router? | A 32-bit Router ID.[1] | A System ID within a Network Entity Title (NET).[4] |
| What describes topology? | Link-state advertisements, carried in Link State Update packets.[1] | Link-State PDUs, usually called LSPs.[4] |
| What represents a shared LAN? | The DR originates a network-LSA; OSPF also elects a BDR.[1] | A DIS originates a pseudonode LSP for each active level.[4] |
Do not turn the table into a claim that either protocol always converges faster. For a design review, ask for measurements on the actual topology, software release and failure scenarios instead.
The biggest difference: areas are not levels
In OSPF, an ABR has attachments to multiple areas and maintains a separate link-state database for each attached area.[1] The backbone distributes routing information between non-backbone areas; it is not simply whichever area contains the most routers.[1]
In IS-IS, an L1/L2 router maintains separate Level 1 and Level 2 databases, with flooding performed within the corresponding scope.[4] Its Level 1 adjacency needs a matching area address, but its Level 2 adjacency can connect to a router with a different area address.[4] There is no requirement to name an IS-IS area “0”; the backbone requirement is a connected Level 2 subdomain.[4]
A simple process flow
Consider this original, illustrative topology—not output from an executed lab:
OSPF:
Access A -- Area 10 -- ABR A -- Area 0 -- ABR B -- Area 20 -- Access B
IS-IS:
Access A -- L1 -- Border A -- L2 -- Border B -- L1 -- Access B
area 49.0001 area 49.0002
Border A and Border B are L1/L2 routers.
For OSPF, Access A learns inter-area reachability through its ABR rather than receiving every remote area's detailed router-LSAs.[1] For a basic IS-IS hierarchy without downward route leaking, an L1 router normally uses a default route toward an attached L1/L2 router for destinations outside its area.[3][4] Level 1 prefixes can be advertised upward into Level 2, while Level 2 prefixes are not simply flooded into the Level 1 database.[3]
Interview trap: an L1/L2 router is comparable to an ABR in purpose, but the analogy does not make the two area models identical.[1][3] When drawing a topology, label OSPF areas on links and IS-IS routing levels on adjacencies; then add each IS-IS router's area address.
How neighbor discovery becomes a usable route
OSPFv2: Hello, database exchange, SPF
OSPF Hellos discover neighbors, and the neighbor state machine decides which relationships should become full adjacencies.[1] For an adjacency that must synchronize, Database Description packets summarize the database, Link State Requests ask for missing information, and Link State Updates carry the requested LSAs.[1] Link State Acknowledgments support reliable flooding, after which the database feeds route calculation.[1]
Hello and bidirectional discovery
-> adjacency decision
-> DBD exchange and missing-LSA requests
-> synchronized database
-> SPF and route calculation
On a broadcast network, two DROther routers can remain at 2-Way without a fault; full adjacencies are built with the DR and BDR.[1] Therefore, “every neighbor must be Full” is not a safe troubleshooting rule.[1]
IS-IS: IIH, LSPs, sequence-number PDUs
IS-IS uses IIH Hellos to discover neighboring intermediate systems and their supported routing levels.[4] LSPs carry topology and reachability information, while Complete and Partial Sequence Number PDUs help identify database differences and recover missing information.[4] On a point-to-point circuit, the routers exchange CSNP summaries when the adjacency comes up and use PSNPs to acknowledge updated LSPs.[4] On a shared LAN, the DIS originates the pseudonode LSP and sends periodic CSNP summaries.[4]
IIH discovery and compatible routing level
-> LSP database comparison using SNPs
-> missing or newer LSP exchange
-> synchronized database at each supported level
-> SPF and route calculation
The two protocols therefore share a workflow, not a packet-for-packet translation. Avoid calling an IS-IS CSNP “an OSPF DBD with another name.”
Practical troubleshooting checklist
Use this as a suggested investigation order, not as permission to change a production routing domain during an interview exercise.
| Symptom | OSPF checks | IS-IS checks |
|---|---|---|
| No neighbor | Confirm interface activation, link addressing, Area ID, Hello/Dead parameters and authentication.[1] | Confirm interface activation, NET/System ID, compatible circuit level and authentication.[4][5] |
| Neighbor does not reach the expected state | Decide whether 2-Way is normal; for ExStart/Exchange failures inspect DBD negotiation and advertised interface MTU.[1] | Inspect adjacency state and per-interface level settings; L1 needs a shared area address.[4][5] |
| Adjacency exists but a remote prefix is missing | Check the relevant area database, inter-area advertisements and backbone connectivity.[1] | Check the correct level's database, attached/default behavior and route-leaking policy.[3][4] |
| Database looks correct but traffic fails | Continue with the selected route, next hop and forwarding-table entry. | Continue with the selected route, next hop and forwarding-table entry.[5] |
For a Junos IS-IS investigation, these read-only commands expose the interface settings, adjacencies and advertised database information.[3][5]
show isis interface detail
show isis adjacency detail
show isis database detail
These are verification commands, not a complete configuration or a claim of live router testing. Match command syntax and available detail to the installed release.
Which protocol should you choose?
My recommendation is to make the choice operational, not tribal. Score the candidate platforms against the features you actually need, the team's troubleshooting skills, automation support and measured recovery behavior.
A reasonable acceptance checklist is:
- Draw the expected adjacency and area/level matrix before configuration.
- Document which prefixes each area or level should contain.
- Test a failed core link and a failed border router.
- Confirm both routing-table selection and actual forwarding recovery.
- Compare the measured outage with the service requirement.
- Avoid a migration justified only by “IS-IS is always faster” or “OSPF cannot scale.”
IS-IS explicitly supports extension to multiple network-layer protocols, but a platform's supported features still depend on its software and hardware.[4] That is a reason to verify the implementation, not a reason to assume every IS-IS deployment is automatically the better design.
A concise interview answer
“Both are link-state IGPs that calculate shortest paths from a topology database.[1][2] OSPF organizes the network around areas and Area 0, with ABRs connecting areas.[1] IS-IS uses L1 within an area and a connected L2 backbone, with L1/L2 routers joining the two scopes.[3][4] A useful operational difference is that IS-IS L2 neighbors do not need matching area addresses, whereas ordinary OSPF neighbors must agree on the link's Area ID.[1][4] I would choose between them based on platform support, operational experience and tested failure recovery.”
Common pitfalls and summary
- Confusing an area with a level: an IS-IS L1/L2 router is not simply an OSPF ABR with different CLI syntax.[1][3]
- Treating an adjacency as end-to-end proof: inspect route selection and the forwarding table as well.[5]
- Confusing two meanings of LSP: here it means IS-IS Link-State PDU, not an MPLS label-switched path.[4]
- Memorizing only protocol names: practice explaining where the databases differ and how a remote prefix reaches the routing table.
The key distinction to remember is OSPF areas and ABRs versus IS-IS routing levels and L1/L2 routers.[1][3]
Related reading
- IS-IS in Service Provider Networks: Design and Operations Guide
- OSPF DR/BDR Election Explained
- OSPF Virtual Links Explained
Preparing for a network engineering interview? Draw the two topologies from memory, explain which neighbors can form, and describe the first three checks you would make for a missing route.
Sources
- [1] https://www.rfc-editor.org/rfc/rfc2328
- [2] https://www.rfc-editor.org/rfc/rfc1195
- [3] https://www.juniper.net/documentation/us/en/software/junos/is-is/topics/example/isis-multi-level.html
- [4] https://www.cisco.com/c/en/us/td/docs/ios-xml/ios/iproute_isis/configuration/xe-16/irs-xe-16-book/irs-ovrvw-cf.html
- [5] https://www.juniper.net/documentation/us/en/software/junos/is-is/topics/task/isis-adjacencies-status-introduction.html
Post a Comment