Silvano Rebai - Fotolia
How Ethernet became the world’s networking standard
Ethernet pioneer Bob Metcalfe outlines the technology’s early history, including the rivalry with IBM and what he did to convince people of the value of networks
Like many of his peers, Bob Metcalfe, a young computer science graduate student at Harvard University in 1969, was eyeing a lucrative career. At the time, it was networking amid the growing desire to link up computers that were becoming more powerful.
Realising that the US Department of Defense’s Advanced Research Projects Agency (ARPA) was giving money to computer science departments, Metcalfe decided to work on ARPAnet, which was shifting its focus from being a military tool to connecting academics.
But Harvard did not let him take the lead in connecting the university to ARPAnet. So he took a job at the Massachusetts Institute of Technology and connected that college to ARPAnet and then to the internet.
In the course of that endeavour, Metcalfe learned how to send bits, one at a time down a very long wire. It was this work, which included identifying and fixing bugs in the University of Hawaii’s Aloha network, that eventually earned him a doctorate.
He then moved to Xerox’s Palo Alto Research Center (Parc) and repeated the task, putting Parc on ARPAnet and paving the way for the first version of Ethernet technology.
“I operated at the boundary of hardware and software,” said Metcalfe. “I would build hardware, I would write software, but just the plumbing level. My whole career has been at the plumbing level.”
The journey from Metcalfe’s original idea for Ethernet technology – which marks its 40th anniversary this week – through to its acceptance as a standard took seven years, with the first sketches penned and signed by Metcalfe in May 1972.
Journey from anti-trust to a standard
In January 1979, Metcalfe left Parc with ambitions to start a company. He didn’t know what that would be, just that there was an entrepreneurial fire burning. The following month, he met with Gordon Bell, vice-president of R&D at Digital Equipment Corporation (DEC), the number two computer company in the world at the time.
“Gordon asked me to design an Ethernet for DEC,” said Metcalfe. “I said I couldn’t do that. First, I felt loyal to Xerox and second, I had already designed the best network I knew, so theirs would be second best.
“So, together, we had the idea of writing a letter to Xerox proposing that DEC and Xerox would work together to get their products to be compatible over Ethernet.”
After a chance encounter with Intel at the National Bureau of Standards, DEC, Intel and Xerox began co-operating on Ethernet. But the spectre of breaking anti-trust law meant that the three companies would need to tread carefully. Rather than co-operating to the detriment of competition, the group came up with a different solution.
“That was the moment when the idea of creating an industry standard Ethernet came up,” said Metcalfe. “And you have to remember that in those days, standards were not made, like they are today. IBM would just announce its next product and that would be the standard right away. So, the idea to bring Ethernet to the IEEE [Institute of Electrical and Electronics Engineers] was an important breakthrough idea.”
Read more about enterprise networking
- New generation Fibre Channel doubles performance to 64Gbps with switches and HBAs now available. But what future for Fibre Channel with NVMe-over-fabrics hitting the datacentre?
- Cisco introduces platform for the internet of the future based on $5bn of investment in areas covering silicon, optics and software, bringing the ability to support massively scaled enterprise networks and 5G networks.
- As pressure grows to support bandwidth demands from advanced high-performance computing uses cases, Juniper Networks announces success in 1,300-mile 400GbE network test.
- Building and maintaining an SD-WAN is perceived as a complicated endeavour, and the industry’s messages are not getting through, says a new report.
By then, Metcalfe had founded 3Com with his lawyer and friend Howard Charney and others. Eventually, 3Com was acquired by HP for about $2.7bn in cash.
On 30 September 1980, DEC, Intel and Xerox produced the Ethernet Blue Book – the first documented Ethernet specification that would go to the IEEE. They gave it to the newly formed committee, IEEE 802 – with no dot.
“Just 802 and we thought it would officially standardise Ethernet,” said Metcalfe. “But IBM and General Motors had different ideas. They didn’t like the idea of somebody else making standards for them, so they each submitted a competing proposal – IBM’s Token Ring and General Motors’ Token Bus, and the IEEE then began to consider the three of them.”
Metcalfe was unhappy with this outcome and said that was where the dots came from – 802.3 was for Ethernet, 802.4 was for Token Bus and 802.5 was for the IBM Token Ring. “From then, there was fierce competition for the next 20 years,” he said.
Although Token Ring was a standard, IBM had a habit of sprinkling “software dust” over its implementation, said Metcalfe. That made interoperability with standards-based competitors such as 3Com very challenging.
IBM’s ‘dark, black heart’
Metcalfe said IBM’s “dark, black heart” was intent on pretending to have a standard but shipping a product that was not standard.
“We never sold very much of it [Token Ring] because IBM, whose heart was not in interoperability, had software dust that they sprinkled over their Token Rings,” he said. “And so, whenever we tried to sell a Token Ring to a customer, we couldn’t get it to work.”
Where Ethernet differed from its competitors was that it remained a layer 1 and 2 protocol.
“We assumed that the whole idea of acknowledging things would happen at a higher level with TCP [transmission control protocol], whereas IBM felt obligated to have acknowledgments,” said Metcalfe.
“Namely, that token in their ring would go around acknowledging that the packet that you sent had gotten there. That was a mistake – IBM’s Token Ring was always slower and more expensive than Ethernet.”
Creating and having a standard was one thing, but taking it to market successfully was another. Although the ability to connect computers over a network seems obvious today, it was a tougher sell in the 1980s. The PC revolution was still in its infancy and the broad use of the internet was still decades away.
First users didn’t see the vision
To sell Ethernet to people with PCs, of whom there weren’t many, 3Com developed a three-node kit that cost $3,000. This allowed people to share a printer or disk and send emails among three PCs. Metcalfe said that although many kits were sold, “sending email among three people wasn’t that interesting”.
As customers struggled to understand the usefulness of a network, Metcalfe came up with an answer to this dilemma. He sneaked into Stanford and created a 35mm slide – this was in the days before PowerPoint – showing that the cost of adding nodes to a network was linear, but the value grew exponentially.
That rule, which says that the effect or value of a network is proportional to the square of the number of connected users of the system, is now called Metcalfe’s Law.
“I made six copies of this slide, because that’s how big the salesforce was,” said Metcalfe. “The slide basically said your network is not useful because it’s not big enough. The remedy for that is to buy more of our products. And they believed us.”
Today’s Ethernet standard bears little resemblance to the one created by Metcalfe back in the 1970s. Access speeds have increased from 2.94Mbps to 800Gbps, with the number of connected devices now exceeding the number of people on the planet.
The internet may owe Bob Kahn, Vint Cerf and Tim Berners-Lee a great debt, but Bob Metcalfe’s Ethernet sits beneath all of them as the giant that has allowed them to see further.
