Chapter 2. A Brief History of Networking Standards

This chapter is meant as background material for readers who may not be familiar with all the types of networking standards there are, and the organizations that produce them. Readers who are already familiar with terms like CCITT, OSI, IEEE, IETF, ITU-T, TCP/IP, X.25, and so on can proceed to Chapter 3, Speeds and Feeds ^[1].

^[1] The information in this short chapter is provided for completeness. There is an unwritten rule that all books that have to do with networking must at some point mention the OSI Reference Model.

In practice, there are two kinds of standards, de facto and de jure. The latter are formally created by organizations that are specifically chartered with producing them, and often have the force of law or treaties behind them. The former were perhaps never intended to become standards, but due to popularity or influence became entrenched in the marketplace.^[2] The standards to be discussed next were de jure standards created under the auspices of the international group known as the CCITT.^[3]

^[2] In the context of this chapter, the standards we discuss are generally the de jure variety.

^[3] The acronym stands for "Comité Consultatif International Télégraphique et Téléphonique." The group still exists. The portion of the CCITT that standardized networking protocols and conducted related activities has been relocated to the International Telecommunications Union, Telecommunications Standardization Sector (abbreviated: ITU-T), which is a component of the United Nations.

The original reason for creating them was partly to offer an alternative to an emerging proprietary networking protocol suite, namely IBM's Systems Network Architecture (SNA), which for many years did achieve de facto standard status until it was displaced by TCP/IP. Pre-OSI networking protocols of the era (mid-1970s and onward), such as IBM's proprietary SNA, were very different from the eventual OSI model in many ways, but shared many architectural features in common with OSI's structure, including the number of layers, and their basic functionality. Although there was not a perfect functional alignment between the OSI-RM and IBM's SNA, it seems rather clear that there was some cross-pollination of ideas among the groups involved in the creation of the two protocol stacks.

One key observation about IBM's SNA is that it had a strong design preference for connection-oriented protocols, and it provided for reliable delivery at many layers, whereas the OSI protocol stack usually offered either connection-oriented or connectionless operation for protocols implementing layers above the Physical layer, which is (of course) inherently connection-oriented.

There is no ambiguity in the term de jure. However, there are multiple ways that a standard can be de facto, including the way it was created and the way it is used. For example, a de jure standard could be used in a way that the authors never intended. If that usage becomes popular, that particular use of the standard could be deemed de facto. For example, certain protocols may have been designed with a certain scope in mind, for example, LAN protocols are typically expected to operate within an area that is "local". However, if that technology later evolves such that it is deployed in larger contexts, that might or might not align with the protocol designers' intentions.

Another way that protocols can evolve is exemplified by the way that Ethernet is beginning to see so-called "jumbo" frames being supported by equipment that supports gigabit Ethernet, however, the IEEE 802.3 committee (which is the name of the working group that is managing the evolution of the standards governing what we typically call "Ethernet") does not recognize such an implementation as being in compliance with the standard. In all other ways, an Ethernet product that supports jumbo frames would be interoperable with earlier devices, but mixed networks might exhibit strange behavior, such as one-way traffic flows for large frames, and other bizarre effects.

The only reason we mention these examples is that it is important to understand that there are no subtleties in the term de jure, but the meaning of the term de facto may be context-dependent, and it is possible to have a de facto usage of a de jure standard.