Network Trailblazers: A Conversation with Robert Metcalfe
The Ethernet pioneer and 3Com founder discusses anonymity on the Internet and the Internet's impact on society.
January 03 , 2012
Return with me now to the thrilling days of yesteryear, 1973 to be precise. It wasn't exactly as if dinosaurs were roaming the streets. But in 1973 computers were, for most people, gigantic, mysterious and expensive animals that lived behind glass windows in secured spaces the size of a gymnasium. Here in a strictly controlled environment the machines were administered by a priesthood of mostly men who wore white shirts, narrow black ties, and spoke in strange tongues like FORTRAN and COBOL.
To be fair, green-screened terminals were turning up in the workplace, all connected to the central computer behind the glass. But Steve Jobs and Bill Gates were still in their teens and the idea of actually putting a computer on every desktop was a vision found in the eyes of only a few pioneers. Many of them were at the Palo Alto Research Center of Xerox where, in 1973, a computer named Alto was developed. The Alto was one of the first computers designed to sit on a desk and be used by one person at a time. It may not have been the first personal computer; it was too expensive for most "persons" to afford and was never sold as a commercial product. But the Alto was clearly a progenitor.
I talked recently with Robert Metcalfe, who on May 22, 1973, circulated a memo at PARC titled "Alto Ethernet." Metcalfe was a 25-year-old electrical engineer, working for Xerox while working on his Harvard doctorate. In this memo, the first to use the name "Ethernet," Metcalfe expanded on work done at the University of Hawaii to describe what has become the dominant technology for connecting devices in local area networks. Metcalfe and colleague David Boggs had a functioning prototype working by the end of the year.
Metcalfe says even with the computer on the desktop, he knew there were resources in the building that everyone would want to share, specifically the also-under-development laser printer and the company Internet (actually, ARPANET in 1973) connection. "It ran a page per second and 500 dots per inch," Metcalfe says, "so the existing methods of connecting the printer would be too slow to keep it busy." That led to a specification for a transmission speed which was 10,000 times faster than anything that previously existed, 2.4MB/sec v. 300B/sec. And with computing power on each desk and the need to connect a large number of devices to the network, the specification called for a new kind of network structure.
Says Metcalfe: "We adopted instead of a box centric model of computing we developed an ‘ether' centric view. That is, where the center is not a box but a communications system. And the approach there was to make it completely passive and have all the computations and all the sharing done by the machines. That led to what was called CSMA/CD, Carrier Sense Multiple Access with Collision Detection access method for sharing that ether, which was originally a large coaxial cable running down the corridors."
While the University of Hawaii's AlohaNet, which Metcalfe studied, included both data packets and collision detection, it featured a star and hub typography and was radio based. The Xerox developers thought it important that their network be medium agnostic, hence the name, a reference to the luminiferous aether for which physicists once searched in vain. As Metcalfe noted, the original Ethernet ran on coaxial cable. One of the original prototype boards can be found in the Smithsonian.
Today we are more likely to find our local area networks connected by twisted pair, Cat-5 and Cat-6 wiring. For the highest data rates we have implementations on fiber. And there are still wireless radio connections, although the specifications of such systems as Wi-Fi are much different from the AlohaNet radio net of 1973.
Metcalfe left PARC in 1979 and founded 3Com, a company to make networking hardware. It wasn't until the IBM PC took hold a few years later that demand for commercial Ethernet products made the company profitable. Metcalfe says he was surprised that it took nearly a decade to reach that point. He says if he knew that he might have specified a slower speed originally, one which would have been less expensive to implement. Ethernet is now administered by the IEEE LAN/MAN Standards Committee (IEEE 802.3) and the standard has expanded to permit much greater speeds.
Asked what is the biggest problem facing the networking community today, Metcalfe says it is security. He says there are technical solutions but they often add to the cost of implementation and many customers resist purchasing the more expensive equipment. And, he adds, there is an ideological dispute over the issue of anonymity. See video, "Metcalfe on Net Anonymity":
Metcalfe lists five directions he sees Ethernet--or networking strategies that can trace their roots to Ethernet--moving, or continuing to move, in the years ahead:
- Faster 100GB/sec
- Into the Core Network or backbone (not just in LAN)
- Wireless Wi-Fi
- Between the LAN and Core (Telecast becomes Carrier Ethernet)
- Down to cellphones and embedded microcontrollers (watches, cameras, etc.)
It's all those microcontrollers that have exhausted the address space, which seemed so big in the days when the Internet protocols were developed. "It seemed like a big enough address space at the time," Metcalfe says. The answer, he notes, is IPv6 and he observes that there is a wide variety of bridging technologies designed to ease the pain of transition. "I was writing columns calling it imminent in the ‘90s, and it's still not widely deployed," Metcalfe says. "There's a bunch of benefits in IPv6 and one of them is the address space is much bigger." Metcalfe thinks it will take another decade for IPv6 to replace IPv4 with the two coexisting during the transition.
I asked Metcalfe if he has been surprised at the impact networking has had on society. He said, yes, surprised and gratified, and he continues to be surprised by it daily. See video, "Metcalfe on Net Impact on Society":
Metcalfe received the Grace Murray Hopper Award from the Association for Computing Machinery in 1980 for his contribution to the development of computer networking. He joined Polaris Venture Partners as a General Partner in 2001, and in January of 2011 he began an appointment at the University of Texas at Austin, Cockrell School of Engineering, where he is Professor of Innovation.
The contents or opinions in this feature are independent and do not necessarily represent the views of Cisco. They are offered in an effort to encourage continuing conversations on a broad range of innovative technology subjects. We welcome your comments and engagement.
We welcome the re-use, republication, and distribution of "The Network" content. Please credit us with the following information: Used with the permission of http://thenetwork.cisco.com/.