Open Shortest Path First (OSPF)

A single OSPF network with 50 routers works beautifully. Every router maintains a synchronized LSDB, runs SPF when topology changes, and converges quickly. But what happens at 500 routers? Or 5,000?

The challenges multiply exponentially:

• LSDB size explodes — Each router stores thousands of LSAs, consuming memory
• SPF computation intensifies — Dijkstra's algorithm on massive topologies strains CPU
• Flooding storms emerge — A single link flap triggers network-wide LSA propagation
• Convergence slows — Every change requires SPF recalculation everywhere

OSPF's answer to this scalability challenge is its hierarchical area design—a partitioning mechanism that divides large networks into manageable domains, containing flooding scope and reducing the computational burden on individual routers.

An OSPF area is a logical grouping of routers and networks that share a common Link-State Database. Routers within an area have identical knowledge of that area's topology, but they don't need detailed knowledge of other areas—they only need summary information to reach destinations in remote areas.

The Fundamental Principle:

Areas create flooding boundaries. LSAs that describe internal topology (Types 1 and 2) flood only within their originating area. This means:

• A link failure in Area 1 doesn't trigger SPF recalculation in Area 2
• Routers in Area 2 don't store detailed topology information about Area 1
• Memory, CPU, and bandwidth are all conserved

Area Identification:

Areas are identified by a 32-bit Area ID, traditionally expressed in dotted-decimal notation (like an IP address). Common practices include:

• Area 0 (or 0.0.0.0) — The backbone area (mandatory, special role)
• Area 1, Area 2, etc. — Represented as 0.0.0.1, 0.0.0.2 in full notation
• Large organizations may use departmental codes: Area 10.0.0.0 for Engineering, etc.

What Areas Provide:

Area 0 holds a privileged position in OSPF's hierarchy—it is the backbone through which all inter-area traffic must transit. This isn't an optional design recommendation; it's a fundamental protocol requirement.

The Two-Level Hierarchy:

OSPF implements a strict two-level hierarchy:

1. Level 1 (Areas): Non-backbone areas containing local routers and networks
2. Level 2 (Backbone): Area 0, connecting all other areas

Every non-backbone area must connect directly to Area 0 through at least one Area Border Router. This creates a hub-and-spoke topology at the routing level:

In this topology:

• B1, B2, B3 are ABRs (orange) connecting non-backbone areas to Area 0
• Traffic from Area 1 to Area 3 must transit through Area 0
• Area 2 routers only need to know how to reach their ABR; the backbone handles the rest

Why This Requirement Exists:

The backbone requirement ensures loop-free inter-area routing. Without a single transit area:

• ABRs might advertise conflicting summaries
• Routing loops could form between areas
• Path selection would become unpredictable

By forcing all inter-area traffic through Area 0, OSPF creates a predictable routing topology where paths between areas always traverse the backbone.

Backbone Design Considerations:

• Backbone stability is critical — Since all inter-area traffic crosses Area 0, backbone instability affects the entire OSPF domain
• Backbone should be well-connected — Redundant links and routers prevent single points of failure
• Keep backbone topology simple — Complex backbone topologies increase convergence time
• Consider backbone-only routers — Routers entirely within Area 0 (no non-backbone interfaces) simplify LSDB management

Area Border Routers (ABRs) are the gatekeepers between OSPF areas. They maintain connections to multiple areas (always including Area 0) and are responsible for:

1. Maintaining separate LSDBs — One LSDB per connected area
2. Generating inter-area routes — Creating Type 3 Summary LSAs for routes learned in one area and advertising them into others
3. Route summarization — Optionally condensing many routes into fewer prefixes at area boundaries
4. Filtering — Implementing routing policies between areas

ABR Definition Subtleties:

The RFC 2328 definition of an ABR is a router with interfaces in multiple areas, one of which is Area 0. However, Cisco's implementation traditionally classified any router with interfaces in multiple areas as an ABR—even without Area 0 connectivity. This matters for:

• Virtual link establishment (only ABRs can terminate virtual links)
• LSA generation behavior
• Show command output

How ABRs Advertise Routes:

When an ABR learns about a network in one area (via Type 1 and 2 LSAs), it can advertise reachability to that network in other areas using Type 3 Summary LSAs. Critically:

• The Type 3 LSA contains only the prefix and cost—not the detailed path
• Routers in the destination area don't know the internal topology of the source area
• The ABR becomes the routing nexus; traffic to the other area goes through the ABR

This abstraction is what enables OSPF scaling—routers don't need thousands of LSAs describing remote area topologies, just a handful of Type 3 LSAs summarizing reachable prefixes.

Route summarization (also called route aggregation) is the process of combining multiple specific routes into a single, less-specific route. Performed at ABRs, summarization dramatically reduces routing table size and LSA count in the OSPF domain.

Example:

Area 1 contains these networks:

• 10.1.0.0/24
• 10.1.1.0/24
• 10.1.2.0/24
• 10.1.3.0/24

Without summarization, the ABR generates four Type 3 LSAs—one for each prefix. With summarization, the ABR can advertise a single 10.1.0.0/22 summary, covering all four networks with one LSA.

Summarization Benefits:

Summarization Configuration:

Summarization is configured on ABRs using commands like (example syntax):

router ospf 1
  area 1 range 10.1.0.0 255.255.252.0

This command tells the ABR: 'When advertising routes from Area 1 to other areas, if routes fall within 10.1.0.0/22, advertise only the aggregate instead of individual routes.'

Critical Consideration: Null Route (Discard Route)

When you configure summarization, the ABR automatically installs a Null0 route (discard route) for the summary prefix. This prevents routing loops if traffic arrives for an address within the summary range that doesn't actually exist. Without this, packets for nonexistent addresses would bounce between routers.

Designing for Summarization:

Effective summarization requires hierarchical IP addressing:

1. Allocate contiguous blocks per area — Area 1 gets 10.1.0.0/16, Area 2 gets 10.2.0.0/16, etc.
2. Subdivide within areas consistently — Use VLSM within the area but maintain summarizability at boundaries
3. Plan for growth — Leave room in the address block for future expansion without breaking summaries
4. Document rigorously — Summarization hides details; documentation reveals them

Beyond standard areas and the backbone, OSPF defines special area types that further reduce LSDB size by limiting which LSAs are allowed. These area types are particularly useful for branch offices and remote sites with limited router resources.

The Problem with External Routes:

In networks connected to the Internet or other external routing domains, OSPF may carry thousands of external routes (via Type 5 LSAs). Every router in the domain stores these LSAs—even branch office routers that only need default routes to reach external destinations.

Area Types Overview:

Totally stubby areas block Type 3 except the default route (0.0.0.0/0)

Stub Areas:

A stub area filters out all Type 5 (AS External) LSAs. Instead, the ABR injects a default route (0.0.0.0/0) as a Type 3 LSA. Routers within the stub area use this default for all external destinations.

Requirements:

• All routers in the area must agree (stub flag in Hello packets must match)
• Cannot have an ASBR inside the stub area (no external route injection)
• Virtual links cannot traverse stub areas

Totally Stubby Areas (Cisco Proprietary):

A totally stubby area takes filtering further—blocking both Type 5 and Type 3 LSAs (except the default). Routers only receive:

• Type 1 and 2 LSAs (internal topology)
• A single Type 3 default route from the ABR

This minimizes LSDB to the extreme—perfect for resource-constrained branch routers.

Designing OSPF areas requires balancing multiple considerations: scalability, convergence speed, administrative boundaries, and addressing constraints. These best practices emerge from decades of enterprise deployment experience.

Sizing Guidelines:

Design Principles:

Common Design Patterns:

1. Hub-and-Spoke (Classic Enterprise)

• Area 0 at headquarters/data center
• Each branch is a stub or NSSA area
• ABRs at WAN aggregation points
• Summarization at ABRs

2. Multi-Campus (Large Enterprise)

• Area 0 spans WAN core
• Each campus is one or more areas based on size
• Multiple ABRs per campus for redundancy
• Full mesh in backbone for resilience

3. Data Center (Modern)

• Leaf-spine topology within Area 0
• Minimal or no additional areas
• Relies on modern hardware's SPF performance
• Emphasizes fast convergence over LSDB reduction

Virtual links are OSPF's mechanism for extending Area 0 across a non-backbone transit area. They're a workaround for situations where physical backbone connectivity isn't possible—and should be understood as exactly that: a workaround, not a design feature.

When Virtual Links Are Used:

1. Backbone partition repair — If a link failure splits Area 0, a virtual link through another area can restore connectivity
2. Acquisition integration — Connecting a newly acquired network that can't immediately have direct backbone access
3. Poor initial design remediation — Fixing designs where areas were created without backbone connections

How Virtual Links Work:

A virtual link is configured between two ABRs across a transit area. The transit area must be a regular area (not stub/NSSA). The virtual link creates a logical point-to-point connection belonging to Area 0, making both endpoints backbone routers.

Configuration Example:

! On ABR1 (Router ID 1.1.1.1)
router ospf 1
  area 1 virtual-link 2.2.2.2

! On ABR2 (Router ID 2.2.2.2)
router ospf 1
  area 1 virtual-link 1.1.1.1

The virtual link is identified by the remote ABR's Router ID and the transit area number.

OSPF's hierarchical area design is what transforms a protocol that could struggle at 100 routers into one that scales to thousands. We've explored the complete area architecture—from fundamental concepts to advanced area types. Let's consolidate the key knowledge:

What's Next:

With area hierarchy understood, we'll dive into the specifics of LSA types—the data structures that carry topology information within OSPF. You'll learn the seven primary LSA types, what each describes, where each floods, and how they work together to enable both intra-area and inter-area routing.