What Is a Core Router?
In a small network there is one router and it does everything. In a large network, whether an internet provider’s, a university’s or a big data centre, routers have different jobs, and the ones in the middle that carry the heaviest traffic are called core routers. Here is what a core router does, how it differs from edge and distribution routers, and what good design around it looks like.
What a core router is
A core router sits in the centre, or core, of a network. Its job is to move very large volumes of traffic between the other parts of the network as fast and as reliably as possible. It is not usually where end users connect and it does little else besides forwarding packets. Think of it as a motorway junction: it does not care where an individual car is going in the end, only about pushing huge amounts of traffic to the right exit without a jam.
Core, distribution and edge
Large networks are commonly built in layers, a design known as the hierarchical model:
- Access layer: where end devices connect, through switches on each floor or in each rack.
- Distribution layer: aggregates the access switches, applies policy such as filtering and routing between segments.
- Core layer: the high-speed backbone joining the distribution layers, data centres and the outside world.
An edge router is different again: it sits at the boundary of the network and connects it to other networks, such as an internet provider. Edge routers deal with the variety of the outside world (many routes, filtering, policy), while core routers focus on raw speed inside.
What makes a core router different
- Throughput: core routers forward packets in specialised hardware (custom chips called ASICs or network processors) at very high speeds, with interfaces from 10 up to several hundred gigabits per second.
- Reliability: they are built with redundant power supplies, cooling, control modules and line cards, and support software that lets them keep forwarding while a component is replaced or upgraded.
- Routing capacity: they run an interior routing protocol such as OSPF or IS-IS to reach the other routers in the same network, and depending on the design they may also carry large BGP tables (many providers instead keep BGP at the edge and use MPLS across the core).
- Scalability: modular chassis let a network add capacity by adding line cards rather than replacing the router.
Examples of platforms used in core and backbone roles are the Cisco ASR 9000 Series and Cisco 8000 Series, the Juniper MX and PTX series and the Nokia 7750 service router family. Enterprise and cloud data centres often use high-end switches with routing features instead, for cost and density reasons.
How a core network stays fast and available
Redundant paths
A core is built so no single link or router failure cuts the network in two. Routers are connected in a partial or full mesh, and routing protocols detect a failure and reroute traffic, often in well under a second when combined with fast failure detection such as BFD (Bidirectional Forwarding Detection).
Using several paths at once
With ECMP (equal-cost multi-path routing), a router spreads traffic across multiple equally good paths, which raises capacity and avoids relying on a single link (because traffic is spread by flow, an individual link can still become congested). This is a routing feature, not the same as an application load balancer, which distributes requests among servers.
Quality of service
When links are busy, QoS (quality of service) policies prioritise sensitive traffic such as voice and video over bulk transfers. It is applied consistently across the core so that markings set at the edge are honoured.
Simplicity in the core
A common design principle is to keep the core simple and fast: complex policy, filtering and per-customer rules belong at the edge or distribution layers, so the core can concentrate on forwarding.
Monitoring and security
Because so much depends on core routers, they are monitored closely for interface errors, utilisation, temperature, routing changes and latency, using tools such as SNMP, streaming telemetry and flow records. On the security side, the control plane (the part that talks routing protocols and accepts management logins) is protected from unnecessary traffic and unauthorised access, management access uses strong authentication and is restricted to trusted networks, and software is updated in planned maintenance windows.
Do you need one?
Probably not at home or in a small office, where a single router and a switch are enough. The term matters if you work in an internet provider, a large enterprise or a data centre, or if you are studying for networking certifications where hierarchical design is a standard topic. For the wider picture of the building blocks, see networking devices from hubs and switches to routers and modems.
Common questions
Is a core router the same as a backbone router?
The terms are used almost interchangeably. Both describe high-capacity routers in the central part of a network.
What is the difference between a core router and a core switch?
A core switch is a layer 3 capable switch used at the centre of an enterprise or data-centre network, usually with more ports at a lower cost per port. A core router in a service provider network typically has larger routing tables and richer routing features. The line between them has blurred.
Does the core route between customers and the internet?
Not directly. Customers connect through access and edge equipment; the core carries the aggregated traffic between them and towards other networks.
Keep reading
- Networking Devices: From Hubs and Switches to Routers and Modems — Where core routers fit among other devices.
- OSI and TCP/IP Models: A Beginner’s Guide — The layers routers work at.
- Network Topologies Explained: Bus, Star, Ring and Mesh — How networks are laid out.
