Want to feel old? Ethernet just celebrated its 50th birthday

Everything goes round in cycles, including computer networking… but not always in rings. The most important networking system so far has vanquished all its loopy rivals for 50 years.

Geoff Huston, the chief scientist of the Asia Pacific Network Information Centre (APNIC), posted a rather splendid potted history of Ethernet yesterday to commemorate its 50th birthday.

It's an information-dense essay and we would do it a disservice if we attempted to précis it for you. It spans from Ethernet's origins as a cabled version of the University of Hawaii's pioneering ALOHAnet wireless networking system, through 10base-5 ("Thick Ethernet"), then 10base-2 ("Thin Ethernet" or Cheapernet, which is about when The Reg FOSS Desk entered this business), then 10base-T over unshielded twisted-pair cabling… which turned into 100base-T ("Fast Ethernet") and Gigabit.

The Ethernet networking system was co-invented at Xerox PARC in 1973 by Turing Award winner Bob Metcalfe and David Boggs, who passed away last year at 71.

Over about the next two decades, many of Ethernet's competitors used ring topologies. The Cambridge Ring was arguably the UK's first, which we mentioned in passing 20 years ago. The Cambridge Ring was decommissioned in 1994, replaced with ATM. At one point, ATM was clearly the future, and would replace all other systems… But back in 2015, we noted that Ethernet had successfully vanquished ATM – and as that piece notes, when it reached 10Gb, it also discarded CSMA/CD, the collision-sensing system that was once one of its definining characteristics.

• Broadcom says Nvidia Spectrum-X's 'lossless Ethernet' isn't new
• Turing Award goes to Robert Metcalfe, co-inventor of the Ethernet
• Switch and router sales surge, with 200/400 gigabit Ethernet kit growing fastest

When Ethernet turned 40, The Reg interviewed Bob Metcalfe who pointed out that it had also beaten Token Ring and indeed the earlier, cheaper ARCnet – also a hub-based, ring-topology system. A decade ago, we said that at some point it might become Terabit, and that is closer now.

Earlier this week, we noted that computer networking was 60 years old. The thing is that when you go back that far, it gets difficult to clearly distinguish networking from the simpler serial connections between mainframes and minicomputers and their dumb terminals. The word "modem" was coined in 1958 to describe the devices that linked the IBM mainframe-based SAGE air-defense computers to remote terminals over the telephone system.

The idea of linking computers together, so that they could share data with each other rather than with puny humans, took longer to emerge. Arguably, that dates back to ARPANET, forerunner of the Internet. But the ARPANET didn't run TCP/IP at first. Although the prelimary proposal for what would become TCP/IP, "A Protocol for Packet Network Intercommunication" [PDF], was published in 1974, it wasn't until 1983 that ARPANET switched to Internet Protocol.

Ethernet 2 was already a published standard [PDF] by then, produced by a group known as DIX: Digital Equipment Corporation, Intel and Xerox.

So the two oldest networking standards around, used by everyone and everything today, are Ethernet and TCP/IP, but the ordering of events is not at all what you might think. First, at the end of the 1960s, came ARPANET, which would evolve directly to become the internet. Then very soon afterwards – remarkably, at the beginning of the 1970s – wireless networking, which a few years later evolved into a bus-topology cabled version: Ethernet. Lastly, TCP/IP was first proposed a year later, although the internet didn't adopt it for nearly a decade… which eventually meant that the PC industry switched to it too, nearly a decade after that. ®