For the beginner, embarking on the study of networking can be daunting as it often presents words and concepts once understood suddenly imbued with new meaning. Computer science is a relatively new idea first proposed in 1956 by Louis Fein in a study for Stanford University. The administrators were dubious of what Fein called "computer-related science." A Computer Science Department would officially be established in 1962 by Purdue University, but a "coherent definition" of this new science would take another three years for recognition and two more years before realizing accreditation. More info
The word semantics, a sub-field of linguistics, was first used by Michel Breal in 1883 to describe how words can have different meanings for different people based on experiential and environmental qualities. In 1967 Robert Floyd would be given credit for programming language semantics, although programming is widely attributed to Ada Lovelace who used an algorithm to calculate Benoulli numbers as far back as 1843. Semantically, programming had an enormous lead over networking in adapting language to suit its purposes. More info
As a younger human I considered a network to be one of the big TV stations of the era, or the phone system. A protocol was a code prescribing etiquette, rules of behavior or a diplomatic exchange. A program was a show on TV or an instructional outline at school; programming was a schedule of those shows and, later, coded instructions to be inserted into some device. An algorithm defined the notion of decidability, a switch was something to turn on the lights, a route was a road map identifier, and a plane was a big transport with wings, or something used to smooth wood, or a two dimensional surface that extended to infinity.
A side note on definitions and perspective:
From a conceptual viewpoint, most networking ideas are interchangeable across vendors. Naturally, I lean toward Cisco for resolution and the last word for any differences that may crop up. As an example, trunking with Cisco is quite distinct as compared to some other vendors. My preference is, of course, obvious. More info
When computing and, eventually, networking came into my life and work, I realized that many of the definitions of things I was comfortable with were suddenly changing daily. My perspective of planes would be challenged as well.
It is agreed that three operational planes exist for the management of network traffic.
Processes traffic through a network device (transit traffic)
Destined for the network device or originated by it
Management Plane (sometimes considered a subset of Control):
Monitors device operation and performance (typically utilizing a protocol such as SNMP)
- Decides where traffic is going (i.e. routing protocols, etc.)
- System configuration, management information
- Exchange topological information
- Policing (Management Plane Protection)
- CLI, HTTP, SSH, etc.
- More info
- Often called the Forwarding Plane
- Utilizes the control plane to forward onto the destination
- Utilizes the control plane to make packet drop determinations
Ivan Pepelnjak has a very nice graphic as well as more specific information on SDN:
Grey areas begin to emerge, and here is where to consider perspective.
- user doc to other user through network device = Data Plane
- user to directly connected network device for CLI access = Data to Management
- network device to network device topology update = Data to Control
Certainly this is a simplification, for it is known there are many protocols at play for any packet stream, in any network at any time slice. In networking, definitions that were once trustworthy and dependable are constantly being reevaluated and redefined to suit a present requirement. Understanding planes requires consideration of perspective; however, the perspective is subject to change. The furtherance of Software Defined Networking will insure that, as abstraction and agility promotes automation and there is less reliance on human intervention.