The management plane is a local or remote interface for managing all aspects of a network element including control plane preferences and hardware.
The control plane understands the different paths through the network, and communicates a subset of those paths, based on policy, to the forwarding plane.
The forwarding plane is sometimes software, but often performance-oriented hardware, that focuses on how to process each packet as it is received including updated packet headers, packet queuing, quality of service, filtering, and transferring packets to the selected output interface, based on forwarding information received from the control plane.
IP Networks have traditionally used a “distributed” control plane composed of routing protocols exchanging information between routers in the network. The information is about paths/routes through the network. The goal is to have all the routers have a common decision about the route through the network that a packet should follow on its way from its source to its destination, and ultimately to ensure the packet is delivered to its destination. A subset of this information is communicated to the forwarding plane, based on policy. The forwarding plane uses this information to decide which output port to forward packets to, as they are received.
The IP control plane provides an automated way to maintain, over time, a common view among routers in the network, of what paths/routes are available in the network and/or the topology of the network. From this information, all routers can determine the next router a packet should be forwarded to, even after topology changes occur. As a result of this automation, IP networks can scale to large networks, with many routers, and many alternate paths/routes between sources and destinations.
A common alternative to a distributed control plane is to configure paths/routes through a management interface, a command line interface or a remote interface. Such approaches can work for small subsections of networks where there are not many alternate paths anyway, and where topology changes are rare.
However, large networks, with frequent topology changes, that cross administrative boundaries would be difficult to operate using an approach like this. Large networks increase the complexity for human planners, frequent topology changes would mean frequent human intervention and/or high-latency updates from a centralized program, and crossing administrative boundaries brings up issues of information sharing and control. For all these reasons and more, the Internet uses a distributed control plane.
That said, there are approaches to augmenting distributed control planes with centralized analysis and path calculations that are being explored by the networking industry. Those approaches are mostly focused on optimizing internally within a single network, and not across the entire Internet as a whole. That subject will be covered in a future article on software-defined networking (SDN).