Voices of Ethernet | Rich Seifert

This is part of the video series The Voices of Ethernet

Rich Seifert doesn’t shy from taking the big swing. This is a person who once delivered a presentation to a computer-networks conference titled, “Top Ten List of Stupid Networking Ideas.”

“I always like to make a splash,” he said.

Waves are still rippling from the splashes that Seifert made over the course of his decades in creation and evolution of Ethernet. In 1979 and ’80, while he was with Digital Equipment Corporation (DEC), Seifert worked alongside engineers from Intel and Xerox to cowrite, “The Ethernet, A Local Area Network. Data Link Layer and Physical Layer Specifications,” the seminal document which greatly informed the initial IEEE Project 802 standardization activities. He went on to play instrumental roles also in development of 10 Megabit per second (Mbps) Ethernet, Fast Ethernet and Gigabit Ethernet.

Along the way, Seifert was responsible for the architecture and design of a wide range of network products, as well as a lengthy list of books and papers that introduced or furthered all sorts of interesting ideas across and beyond the ecosystem—“ether-not,” for example, is Seifert’s term for technologies that fail to deliver Ethernet’s unique and differentiating combination of attributes.

“Ethernet started off as a specific speed, a specific cable, a specific packet format … a lot of specifics … We now have how many zillions of different Ethernet media—10BASE-T, 100BASE-T, Gigabit, fiber, various other cables, wireless and personal area networks and whatever?”

The common denominator across the innovations that have flourished from Ethernet’s earliest iterations is “a concept of high-bandwidth networking with as little intelligence as possible in the network,” said Seifert.

“It’s the exact opposite of telephony. Traditional telephone systems have very simple devices—dumb telephones that sit on your desk that have nothing in them but a speaker, a microphone and a ringer—and all of the intelligence is in the network, in the central office and the switches. Ethernet’s the other way around. There’s almost no intelligence in the network. The network is just a pipe; it’s a way to get packets quickly from one place to another. All of the smart stuff—the packetization, the reassembly, the reliability—that’s all in the higher-layer protocols, which is all running in the end stations, the clients and the servers. So we keep the networks simple, which means we keep the networks cheap, which means we can make lots and lots of them.”

Seifert is proud of the technology’s story and his roles in its twists and turns.

“I’m glad I chose to work on the Ethernet as opposed to something else,” he said. “… You never know what project you’re going to work on that’s going to change your life and possibly change the world.”