Static vs Dynamic Routing

Once a network has more than two routers, you have a choice: configure routes manually (static) or run a protocol that builds routing tables automatically (dynamic). Most real-world networks use both.

Static routing

An administrator manually configures routes on each router. The routes don’t change unless someone edits them.

# Linux
sudo ip route add 10.10.0.0/16 via 192.168.1.254

# Cisco IOS
ip route 10.10.0.0 255.255.0.0 192.168.1.254

When static is the right choice

• Small networks (a few routers, few subnets)
• Stub networks (single uplink, no choice of paths)
• Default routes pointing to your ISP
• Override behavior — force specific traffic over a specific path
• Lab environments, home networks

Pros

• Simple, predictable, no protocol overhead
• No CPU/memory cost on routers
• Easy to reason about — what you wrote is what you get
• No security exposure from route advertisements

Cons

• Doesn’t adapt — if a link fails, traffic blackholes until manually fixed
• Doesn’t scale — 100 routers × 100 routes each = 10,000 lines of config
• Easy to make typos that cause outages
• Requires touching every router when topology changes

Dynamic routing

Routers run a protocol that exchanges routing information with neighbors. The protocol computes the best paths and updates routing tables automatically. When a link goes down, the protocol detects it and reroutes within seconds.

The two main families

Interior Gateway Protocols (IGPs)

Run inside one organization. Examples:

• OSPF — link-state, fast convergence, dominant in enterprise
• IS-IS — like OSPF, common at large ISPs and internet backbones
• EIGRP — Cisco proprietary (mostly), hybrid distance-vector + link-state
• RIP — old distance-vector, only for very small or legacy networks

Exterior Gateway Protocols (EGPs)

Run BETWEEN organizations. Only one matters today:

• BGP — what the entire internet runs on. Each ISP and large enterprise has an Autonomous System (AS) and uses BGP to peer with others.

When dynamic is the right choice

• Networks with multiple paths to the same destination
• Networks that need to survive link failures automatically
• Anything spanning multiple sites or data centers
• ISPs, cloud providers, large enterprises

Pros

• Adapts to failures within seconds
• Scales to thousands of routers
• One config change propagates through the network
• Multi-path load balancing automatically

Cons

• Protocol complexity — OSPF areas, BGP route maps, etc. take time to learn
• Convergence delays — even dynamic protocols take seconds to react
• Misconfiguration can cause spreading outages (BGP especially)
• Routing loops are possible if not configured carefully

Distance-vector vs link-state

Two algorithm families behind dynamic routing:

Distance-vector (RIP, BGP)

“Tell my neighbors what I know.” Each router only knows about its neighbors and what they tell it. Slow to converge, simple to implement.

Link-state (OSPF, IS-IS)

“Flood my neighbor info to everyone, then everyone computes the topology themselves.” Each router builds a complete map of the network. Faster convergence, more CPU.

Real-world: most networks combine both

Typical setup:

• Default route on every device → static, points to local gateway
• Internal routes within a site → OSPF
• Routes between sites/data centers → BGP
• Override routes for specific traffic → static, more specific than dynamic learned routes

Cloud routing

In AWS/GCP/Azure VPCs, you don’t run routing protocols. The cloud provider does it for you. You configure “route tables” — which look static but are managed by the cloud’s underlying infrastructure.

What to learn next

BGP — the protocol that holds the internet together. The only EGP that matters. Up next.