Anarchy on the Internet is what makes it work cost-effectively. Cisco Systems pioneered much of the Internet’s plumbing during the early 2000s, placing its bet on IP-based routed traffic instead of on the cumbersome ATM-infused network designs of the 1990s.
Researchers discovered, in 2002, that congestion of IP traffic is not such a bad thing. They proved, according to Brian Christian and Tom Griffiths [1], “that the ‘selfish routing’ approach has a price of anarchy that’s a mere 4/3. That is, a free-for-all is only 33% worse than perfect top-down coordination . . . Selfish routing’s low price of anarchy may explain, for instance, why the Internet works as well as it does without any central authority managing the routing of individual packets. Even if such coordination were possible, it wouldn’t add very much.”
Indeed, centralized ATM switches, and eventually centralized voice-based (Class 5) switching platforms became obsolete as IP-based traffic and Ethernet transport [2] overtook them during the dawn of the new millennium.
Distributed communications, owing to its cost-effectiveness, came to dominate telecommunications network designs of the 2000s. Ethernet quickly migrated from the office in the form of Local Area Networks (LANs) to the enterprise across Wide Area Networks (WANs). Eventually, metropolitan and global connections, linked by optical transport systems, completed Ethernet’s dominance. The ability to connect every IP-enabled device to every other such device achieved wide appeal. Ethernet and Fast Ethernet were soon dwarfed by Gigabit Ethernet (GbE) and then 10 GbE connections by 2002.
Researchers discovered, in 2002, that congestion of IP traffic is not such a bad thing. They proved, according to Brian Christian and Tom Griffiths [1], “that the ‘selfish routing’ approach has a price of anarchy that’s a mere 4/3. That is, a free-for-all is only 33% worse than perfect top-down coordination . . . Selfish routing’s low price of anarchy may explain, for instance, why the Internet works as well as it does without any central authority managing the routing of individual packets. Even if such coordination were possible, it wouldn’t add very much.”
Indeed, centralized ATM switches, and eventually centralized voice-based (Class 5) switching platforms became obsolete as IP-based traffic and Ethernet transport [2] overtook them during the dawn of the new millennium.
Distributed communications, owing to its cost-effectiveness, came to dominate telecommunications network designs of the 2000s. Ethernet quickly migrated from the office in the form of Local Area Networks (LANs) to the enterprise across Wide Area Networks (WANs). Eventually, metropolitan and global connections, linked by optical transport systems, completed Ethernet’s dominance. The ability to connect every IP-enabled device to every other such device achieved wide appeal. Ethernet and Fast Ethernet were soon dwarfed by Gigabit Ethernet (GbE) and then 10 GbE connections by 2002.
Cogent, a competitive local exchange carrier, was a pioneer in offering Ethernet-based services across its networks. Today, Cogent boasts a global footprint across 198 markets throughout 41 countries in North America, Europe and Asia, with over 57,200 route miles of long-haul fiber and over 30,190 miles of metropolitan fiber. Image courtesy Cisco, circa 2000.
One of the technology pioneers in delivering these high-speed connections was Cerent and its acquired engineering team within Cisco. During mid-1999, a team led by Bob Bortolotto at Cerent, engineered the E-series “Ethernet-over-optical” option on the Cerent 454. New service provider players, such as Cogent, provided input regarding the Ethernet switching capabilities on Cerent’s optical transport platform.
Bob was intrigued by the notion of “packet-over-SONET” being hyped during the mid-1990s by Vinod Khosla, a Silicon Valley venture capitalist. Bob recalls talking about packet-over-SONET with Vinod: “Vinod was part of the interview process and so I met with him. I just liked everything that I heard. I didn’t know anything about SONET or transport. I knew about the access side, but not the transport side of the business. That was all new to me. But the idea that they were trying to merge Ethernet switching and routing into this platform intrigued me. I was perfect for it because that was my background. I had the telecom background, the data networking background and it seemed like it was a great fit. So I accepted the job in the summer of 1997.”
Ethernet, Fast Ethernet (100BaseT), and Gigabit Ethernet Arrive
This packet-over-SONET solution in the form of the early E-series option morphed into a Cisco-influenced improvement of “Ethernet-over-optical” known as the ‘454’s’ G-series option. Compared to one of Cisco’s major competitors during the early 2000s, Nortel found it difficult to keep up with the innovation introduced by Cerent-Cisco. The performance graph shown below is an indication of Nortel playing catch up in the emerging IP-based networks, an established company often held back by the legacy of its aging circuit switched-based networks.
Ethernet, Fast Ethernet (100BaseT), and Gigabit Ethernet Arrive
This packet-over-SONET solution in the form of the early E-series option morphed into a Cisco-influenced improvement of “Ethernet-over-optical” known as the ‘454’s’ G-series option. Compared to one of Cisco’s major competitors during the early 2000s, Nortel found it difficult to keep up with the innovation introduced by Cerent-Cisco. The performance graph shown below is an indication of Nortel playing catch up in the emerging IP-based networks, an established company often held back by the legacy of its aging circuit switched-based networks.
The statistics bore out this trend. “Worldwide revenues for intelligent optical network hardware hit $11.4 billion in 2001, a 67 percent growth over 2000,” reported Fiber Optics News [3]. This trade publication added that revenues “are projected to double to $23 billion in 2005, according to Infonetics Research.”
While the overall optical hardware market was actually declining because of reductions in SONET-SDH transport spending (estimated by some at $25–$30 billion in 2001), “the intelligent optical hardware market is on the rise.” The story continued, “This is for two reasons: optical equipment spending will likely being rebounding in North America by late 2003, and, more importantly, many historically ‘non-intelligent’ products such as DCS, old SONET-SDH, and closed DWDM are being superseded by intelligent products incorporating these technologies.” These technologies were presented as MSPPs to the market by Cerent, and then Cisco after the Cerent acquisition.
And Today Ethernet Continues to Thrive
Last year, 10Gb Ethernet switch port shipments “grew a robust 24.7% year over year with over 9 million ports shipped in 2Q16” alone, according to IDC [4] . . . 40Gb Ethernet shipments “reached a record 1.9 million ports, for an increase of 97.5% year over year. 10Gb and 40Gb Ethernet have been the primary drivers of the Ethernet switch market . . . in 2016 “as 25/100Gb Ethernet start to emerge and will see volumes pick up.”
From thousands of Ethernet ports to millions of ports in shipments over almost two decades makes for a $6 billion annual Ethernet switch market today.
[1] Algorithms to Live By: The Computer Science of Human Decisions, Brian Christian and Tom Griffiths, 2016, Henry Holt, New York, pp.236–237.
[2] Bob Metcalfe, of IBM’s Palo Alto Research Center (PARC), simply sought an easy way to connect his computer to a printer. As a result, he shares four patents for the introduction of Ethernet. According to a Mad Science entry, “He and PARC colleague David Boggs published the concept in a 1976 paper, Ethernet: Distributed Packet-Switching for LANs. Metcalfe convinced funders Xerox, DEC, and Intel to let Ethernet become an open networking standard. It eventually supplanted competing technologies like IBM’s Token Ring and General Motors’ Token Bus [and Apple’s Apple Talk] as the predominant standard for local area networks [and two decades later as the standard for Wide Area Networks or WANs]."
Source: Mad Science, Editor Randy Alfred, 2012, Little Brown and Company, NY, p.144.
[3] Fiber Optics News, Intelligent Optical Network hardware Market Hits $11.4B in 2001, March 11, 2002, p.8.
[4] IDC's Worldwide Quarterly Ethernet Switch and Router Trackers Show Strengthening Ethernet Switch Market and Softer Router Revenues, Business Wire, September 16, 2016.