Topic 20

Static vs Dynamic Routing

Routing

Routes reach a routing table two ways. A static route is one you type in by hand; it sits there unchanged until you edit it again, indifferent to whether the link it points at is up or dead. A dynamic route is learned by a routing protocol that talks to neighboring routers, discovers the topology, and rewrites the table on its own as links appear and disappear. Static is predictable and dumb; dynamic is adaptive and complex.

Real networks are not a choice between the two — they blend them. Static routes anchor the simple, stable edges: a stub site with one uplink does not need a protocol to know everything goes out that one link. Dynamic protocols run in the core and anywhere with redundant paths, where the whole value is reacting to failure faster than any human could. The skill is knowing which tool each part of the network has earned.

Statictyped by hand

Control and simplicity. Predictable, protocol-free, zero CPU — but blind to failure: a dead next-hop stays in the table and black-holes traffic until you edit it.

Dynamiclearned by a protocol

Adaptivity and convergence. OSPF/BGP track the topology and reroute around failures in sub-seconds — at the cost of neighbors, state, and CPU spent converging.

Static Routes

A static route is explicit and protocol-free: no neighbors, no advertisements, no CPU spent on convergence. You state a prefix and a next-hop, and the router obeys exactly that. For a stub network — a branch office with a single path to the rest of the world — this is the right answer. There is nothing to compute because there is no choice to make, and a static default route out the one uplink is simpler and more predictable than any protocol.

The cost is rigidity. A plain static route has no idea whether its next-hop is reachable; if the link dies, the route stays in the table and traffic pours into a black hole. Static routing does not fail over by itself. Getting failover requires bolting on something extra — a tracked object, BFD, or a floating static route with a higher administrative distance that only activates when the primary withdraws — because the static route alone will never notice the failure.

# a static default route plus a more-specific static for one subnet
ip route add default via 203.0.113.1
ip route add 10.20.0.0/16 via 10.0.0.2

# a "floating" backup: higher metric, only used if the primary withdraws
ip route add default via 198.51.100.1 metric 200

Dynamic Routing Protocols

Dynamic protocols fall into three families by how they learn. Distance-vector protocols (RIP, EIGRP) tell their neighbors "I can reach X at cost N" and trust the rumor — each router knows only what its neighbors report, never the full map. Link-state protocols (OSPF, IS-IS) flood every router's local link information to everyone, so all routers build one identical map and each computes paths itself. Path-vector protocols (BGP) carry the entire list of networks a route traversed, trading shortest-path math for policy control.

The family determines the failure mode. Distance-vector's rumor model is simple but prone to slow convergence and counting-to-infinity loops without hacks like split horizon. Link-state converges fast and avoids loops because everyone shares the same map, at the cost of more memory and CPU to hold and recompute it. Path-vector scales to the whole internet precisely because it does not try to find the shortest path — it follows policy, and the AS-path it records makes loops impossible to miss.

Convergence

Convergence is the time between a topology change and the moment every router agrees on the new set of best paths again. During that window the network is inconsistent: some routers know about the failed link, others don't, and packets can loop, drop, or take a stale path until everyone catches up. Convergence is the single number that most defines a routing protocol's quality — link-state designs converge in well under a second, while older distance-vector setups could take tens of seconds.

The enemy of convergence is instability. A link that flaps — up, down, up, down — forces the whole protocol to recompute on every transition, and the recalculation traffic can itself overwhelm the routers in a convergence storm. This is why protocols add dampening: deliberately ignoring a route that has changed too often recently, trading a slower reaction to a genuinely flapping link for protection against a storm that would take everything down.

Administrative Distance and Metrics

When two sources offer a route to the same prefix at the same prefix length, the router needs a tie-breaker. Administrative distance ranks the sources, with lower winning: a connected route beats a static, which beats eBGP, which beats OSPF, which beats RIP. The values are a vendor convention, not a standard — on Cisco the defaults are connected 0, static 1, eBGP 20, OSPF 110, RIP 120, and other platforms pick their own. It is how a router decides which protocol to believe when several claim the same destination — a static route, by default, is trusted over anything a protocol learned.

The metric is the within-protocol cost, used only after administrative distance has already chosen a protocol. OSPF's metric is link cost (derived from bandwidth); RIP's is hop count, capped at 15. A floating static route exploits AD directly: give the backup a higher administrative distance than the primary, and it stays out of the table until the primary disappears, at which point it is the only candidate left and takes over.

Static vs Dynamic

Static routing gives you total control and zero protocol overhead — no neighbors, no CPU, no surprises. Choose it for stub networks, single-homed sites, and any path that genuinely never changes. The price is no automatic failover: when a link dies, you (or an external tracker) must react.

Dynamic routing adapts on its own, rerouting around failures in seconds and scaling to topologies no human could track by hand. Choose it for any network with redundant paths or more than a handful of routers. The price is complexity and resource cost — and a misconfiguration can ripple far beyond where you made it.

Common Mistakes

A plain static route with no failover, pointing at a link that later dies. The route stays installed, the next-hop is dead, and traffic black-holes — static routing never notices a failure on its own.
Running a full link-state IGP on a tiny three-router stub. You pay the memory, CPU, and adjacency complexity of a protocol to solve a problem a single static default route already solved more reliably.
Misconfiguring a metric or administrative distance so traffic takes the long way. A backup link with a more attractive metric than the primary quietly becomes the preferred path, adding latency nobody asked for.
Leaving a flapping link undampened, so every up/down transition triggers a network-wide recompute. The recalculation traffic compounds into a convergence storm that destabilizes routers far from the actual fault.
Assuming a longer prefix loses to a lower administrative distance. AD only breaks ties among routes of equal prefix length; a more specific route still wins first, regardless of which protocol or AD produced it.

Best Practices

Use a static default route for single-homed stub sites instead of a routing protocol — it is more predictable, costs nothing to run, and removes an entire class of misconfiguration.
Attach reachability tracking (BFD, IP SLA, or an object tracker) to any static route that must fail over, since a bare static route cannot detect its own next-hop going down.
Build a floating static backup by giving it a higher administrative distance than the primary, so it stays dormant until the primary withdraws and then takes over automatically.
Enable route dampening or interface debouncing on links prone to flapping, trading a slightly slower reaction for protection against a convergence storm that would hit the whole domain.
Reserve dynamic routing for cores and any topology with redundant paths, where automatic reconvergence in under a second is worth the added CPU, memory, and operational complexity.

Comparable conceptsOSPF (link-state, next)BGP (path-vector, next)

Knowledge Check

For a branch office with a single uplink to the rest of the network, why is a static default route usually the better choice?

There is only one path, so a protocol adds overhead and complexity without any benefit
A static route fails over to a backup path faster than any dynamic protocol ever could
Static routes always have a better metric than learned routes
A static route automatically detects when its uplink goes down

What does "convergence" describe in a dynamic routing protocol?

The time after a topology change until every router agrees on the new best paths
The merging of every router's table into one shared master copy of all the routes
The compression of routing advertisements to reduce protocol traffic
The selection of one route when several prefixes are equally specific

Two routes to the same prefix, equally specific, come from OSPF (AD 110) and a static route (AD 1). Which is installed, and why?

The static route, because its lower administrative distance wins the tie
The OSPF route, because a higher administrative distance is always preferred
The OSPF route, because dynamic protocols always override static ones
Both, split evenly, since the two routes have the same prefix length

You got correct