A stub router is a router positioned at the edge of a network that only handles traffic for its own directly connected networks, never serving as a pass-through path for traffic between other routers. It’s most commonly associated with EIGRP (a Cisco routing protocol), though the concept also appears in OSPF with some important differences. Stub routers are designed for small branch offices and network endpoints where there’s no reason for backbone traffic to flow through.
How a Stub Router Works
In a typical network, when a router loses a route to a destination, it sends out queries to its neighbors asking if they know an alternative path. Those neighbors then query their own neighbors, and so on. In large networks with hundreds of branch offices, this cascading wave of queries can consume significant bandwidth and CPU resources, sometimes called a “query storm.”
A stub router solves this by announcing itself as a dead end. When you configure a router as a stub, it adds a flag to the periodic hello messages it sends to its neighbors. When a hub router sees that flag, it knows not to bother asking the stub for alternative routes, because a stub router at the edge of the network wouldn’t have backup paths to offer anyway. The stub router only advertises the networks directly attached to it, and its neighbors suppress all queries to it.
This does two things simultaneously: it stops unnecessary query traffic from reaching branch office routers over potentially slow links, and it prevents the network from trying to route backbone traffic through a small branch office connection that can’t handle it.
Where Stub Routers Fit in Network Design
Stub routers belong in hub-and-spoke topologies, which are extremely common in enterprise networks. Picture a corporate headquarters (the hub) connected to dozens or hundreds of branch offices (the spokes). Each branch office router connects back to headquarters but has no direct connections to other branches.
In this design, a branch office router has no business carrying traffic between headquarters and another branch, or between two other parts of the backbone. Its only job is to route traffic for its own local network. Configuring it as a stub formalizes that role. If a backbone link fails somewhere in the core network, traffic won’t be rerouted through a branch office’s low-speed connection. The branch router stays focused on its own local traffic, and the rest of the network converges faster because it doesn’t waste time querying endpoints that can’t help.
What a Stub Router Advertises
By default, an EIGRP stub router advertises only two types of routes: connected routes (the networks physically attached to it) and summary routes. It will not advertise routes it learned from other neighbors, even if the configuration would otherwise allow it. This is a key point that trips people up during configuration.
You can adjust what a stub router shares using several keyword options:
- Connected: advertises the networks directly attached to the router’s interfaces
- Static: advertises static routes that have been manually configured
- Summary: advertises summarized route information
- Receive-only: the router advertises nothing at all, making it essentially invisible to the rest of the network from a routing perspective, like an end device
A common configuration mistake involves static routes. If your branch office has a static route that other parts of the network need to know about, the default stub configuration won’t advertise it. You need to explicitly add the “static” keyword. And here’s the gotcha: if you add “static” but forget to also include “connected,” the router will only advertise the static routes and drop the connected ones. You have to list every type you want.
EIGRP Stub vs. OSPF Stub Areas
The term “stub” appears in both EIGRP and OSPF, but the two features solve different problems. In EIGRP, the stub feature is about convergence speed. It tells neighbors not to query the stub router when searching for lost routes, which reduces query storms and speeds up how quickly the network adapts to changes.
In OSPF, a stub area is about reducing the size of the routing table. Routers inside an OSPF stub area don’t receive detailed external route information. Instead, they get a single default route that points toward the rest of the network. This conserves memory and processing power on smaller routers that don’t need a complete map of every destination outside their area.
OSPF also has a separate concept described in RFC 3137 where a router advertises itself as a stub to prevent transit traffic. It does this by setting the cost of all its non-local links to the maximum possible value (65,535), making other routers avoid it as a path while still being able to reach the router’s own addresses. This is useful during maintenance windows, when a router is overloaded, or when gracefully adding or removing a router from the network.
Performance Benefits
The practical gains from stub routing are most noticeable in networks with many branch offices. Each suppressed query is one less packet traveling across a potentially slow WAN link, and one less computation the branch router’s CPU has to perform. Multiply that across hundreds of spokes during a network event, and the bandwidth and processing savings become substantial.
Convergence time improves because the hub routers don’t have to wait for responses from dozens of stub routers that were never going to have useful alternative paths. The network reaches a stable state faster after a topology change, which means less disruption for users.
Low-speed links to remote offices also get protection. Without stub routing, a backbone failure could cause traffic meant for high-bandwidth core links to be rerouted through a branch office’s modest connection, overwhelming it. With stub routing in place, the branch link is reserved exclusively for that branch’s own traffic.