The Internet changed three fundamental things about networking

In the first article of this series on the history of managed services, we saw how a number of major accidents of history - including the rise of the Internet and the breaking up of AT&T - eventually led to the global Internet we know today. But the subversive effect of the Internet goes deeper than that: it changed three major things about computer networking.

Firstly, it has changed applications and how they communicate. One of the primary examples is e-mail. Right up until the mid-1990s, there were a number of popular competing standards for e-mail: internal proprietary ones for use within single-vendor environments, open standards such as X.400 which nonetheless were painful to implement, and open standards like SMTP which were easy to implement (from scratch if required) and that would eventually came to dominate as a result. Today very few networked applications outside some small niches speak anything other than Internet Protocol to each other. Networking, be it over a local LAN or the global Internet, is a solved problem and little to no investment needs to be made to network an application. Whereas once applications (and even games) were single user or restricted to local networks only, that is now the exception. Everything else is networked.


The second area the Internet has simplified is the network itself, as made up of the protocols that run over some kind of physical hardware. Twenty years ago, effective communications meant choosing the right protocol and choosing the right network infrastructure on which to run it. Pitched battles were fought between vendors about wide area network standards in particular: which was more efficient, cost-effective and manageable depended on who was believed at the time. Similarly, there were arguments about the long term future of network architectures like ’s Systems Network Architecture (SNA), Novell’s IPX (for local area networks), and some of the dial-up protocols that connected modems to the telephone network, arguments that now seem quaint with hindsight. This is not to say that alternatives  were worse. SNA in particular is a very resilient architecture that can recover from loss of data, avoid network congestion automatically, identify failures quickly, and recover from errors without user intervention. Many ATMs in this country still use it to communicate with the owner bank’s data centre.

However, the Internet made most of the arguments go away. In 1993 the Internet carried only 1% of the information flowing through two-way telecommunication. By 2000 this was 51% (helped by the boom) and by 2007 it was up to 97%, helped by the fact that almost all voice calls are routed over the Internet and because vendors no longer waste money building global networks than run proprietary protocols.

Even local area networks now have Internet protocol (with private addresses) almost universally. Speaking to the outside world is done through a gateway, something that is so completely transparent to everyone that it’s hard to believe that it used to be so limited and restricted in the past. And because IP is a standard and computers communicating with it can speak to any others without worrying about compatibility or vendor lock-in, a very real problem in the days of multiple networking protocols. Indeed it spawned a bunch of solution providers who could translate between different dialects.


The final, and perhaps most subversive thing of all changed by the Internet, was the power of the end user. I don’t mean the consumerisation of IT that is happening now but something that happened earlier, mainly due to a deliberate design decision by the original architects of the Net. Unlike many other networks of twenty years ago and earlier, the Internet was a “dumb” network. The intelligence was to be found at the ends of the network: in its users and in software that connected to it. Armed with a standards document and some software skill, people could - and did and still do - write programs that changed the world, unencumbered by restrictions imposed on them by the owners of the network. Proprietary networks were draconian in what they permitted, the Internet was not. The barrier to entry for a globally networked application used to be in the hundreds of millions of dollars. Now a popular application can be written in a few hours and be instantly available to anyone in the world with a connection.