Metcalfe had come back to PARC to pitch an eventcelebrating the 40th anniversary of Ethernet and innovation in general. The event will include an innovation contest and a fundraiser for STEM education.
The power of open standards was one of the big lessons Metcalfe took away from the fights to define and establish Ethernet. At the time his archrival was IBM and its Systems Network Architecture, a stew of generally IBM proprietary technologies.
Metcalfe notes that Ethernet was invented at a time when IBM and AT&T were still giants that dominated the computer and communications landscape. “I think Ethernet took IBM by surprised and they never really got it, and I think it was what eventually brought IBM down” from industry dominance, Metcalfe said.
“In the mid ‘80’s IBM lost its power to make inexorable standards by simply announcing them, but it recovered and it’s now one of my favorite startups,” he quipped.
The other big lesson for Metcalfe is ‘build it and they will come.’
“Our original goal with Ethernet was to send a seven-bit ASCII character across and back in seven seconds,” Metcalfe said. “Internet traffic now consists of 86 percent video, but when we were developing the Arpanet we were not anticipating YouTube."
Expect the unexpected, and plow through critics who say we don’t need more bandwidth, said Metcalfe. He saw Ethernet rise from a few measly kilobits a second to 100 Gbits/s today.
“From the bandwidth viewpoint every single time there’s a chorus saying we have enough bandwidth and every single time so far there has been elasticity of demand,” he said.
Should also point out that the triumvirate of Digital Equipment Corp. (DEC)--remember those guys, Intel, and Xerox also had a hand in creating the 10-Mbit Ethernet standard that prevailed over the IBM token ring network and others. Funny how much history gets lost in time.
IMHO, the token ring is a superior protocol! It is deterministic, one could allocate bandwidth and priority to a given node, and many other benefits. The problem was marketing ant the greediness and over protection of IBM.
And look at the current Ethernet! Just a point-to-point, switched network, with a single point of failure.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.