How the Internet got fast

Memory lane

Boston College was founded in 1863 to provide education for children of Irish-Catholic immigrants. After outgrowing its original location in Boston’s South End, the college was relocated during the early part of the 20th Century to the Chestnut Hill village of Newton, where in 1909 construction began on Gasson Hall, the first of a series of signature, Gothic-style buildings. Those buildings and their stern exteriors, built from stones plucked from farmland nearby, remain the physical centerpieces of a campus that serves close to 13,000 students today.

Participants in the modern era of cable telecommunications engineering may recognize Boston College for more than its impressive educational and architectural heritage. In the early 1990s the college played an unlikely role in the birth of what has become an astonishingly influential and pervasive communications technology.

With little fanfare, the Boston cable company Continental Cablevision had attached to the transmission lines that snaked their way to the school a number of rectangular devices roughly the size of the Boston phone directory. Stuffed with custom-designed microprocessor chips and rigged with coaxial cable connectors, the boxes were manufactured by LANCity, a Massachusetts upstart bent on convincing cable companies their networks could accommodate a different type of electronic signal than the 6 MHz television channels that normally flowed across cable wires.

As is often the case with emerging communications technologies, some of the first people to make use of the new devices were college students. Originally, about 100 Boston College students used them to connect to a wide-area data network that allowed them to peer into electronic computer databases and exchange typed messages. At the time, those functions weren’t particularly novel. Early adopters had been tapping into the Internet, and to packaged online services like Compuserve, for years. What made the Continental experiment stand out was the speed and ease of mining these data sources. Unlike sluggish baud-level modems of the era, the LANCity modems whisked data to users at then-astonishing rates of greater than one megabit per second – a throughput level then familiar only to users of the fleetest corporate data networks. The students who were interacting with the trial network were among the first in the world to connect to online databanks and to the Internet using a high-speed cable modem. By September 1994, Continental had outfitted nearly every Boston College dorm room with the gadgets.

The Boston College trial wasn’t the only effort under way at the time to test whether cable television networks originally designed to shuttle video signals could withstand the peculiarities of digital data packets. There were important trials under way in Castro Valley, Calif., involving Viacom Cable, and in the Philadelphia suburb of Lower Merion, Penn., involving Comcast Cable. But the endorsement of the Boston project by one of cable’s most-respected operators, Continental, made the effort stand out.

So did its preliminary conclusions, which included a realization that, with a deft level of care applied to the touchy upstream data path, cable television systems indeed were capable of accommodating a new type of signal stream. As similar findings popped up from other locales, the scent of big opportunity was in the air. In 1995, a Silicon Valley venture capitalist, John Doerr, working with cable’s Tele-Communications Inc., created a new company, @Home Corp., to leverage the emerging cable-data business opportunity. More important, as it turned out, cable companies began to share findings about their data-delivery endeavors by forming a highly secretive cabal called Multimedia Cable Network Systems. Its aim: to produce a do-it-all microprocessor for use in cable modems. After much wrangling, the work of that private venture was absorbed by the not-for-profit cable industry research and development center, CableLabs. It was CableLabs that then devised a set of mutual interoperability specifications for the budding cable modem marketplace – a project that now goes by the familiar acronym of DOCSIS.

In August of this year, about 12 years after college students in Boston began tinkering with cable modems, CableLabs released a new set of specifications for version 3.0 of DOCSIS. The new documentation describes approaches for data delivery that could result in downstream speeds of 160 Mbps.

It’s instructive to note that although digital subscriber-line telephone technology dominates the world market for high-speed Internet access, in the U.S. it’s DOCSIS that still rules, with cable companies commanding about 56 percent of the high-speed data market, says Leichtman Research Inc. The lesson? Building the right foundation can support tremendous progress through subsequent generations. Just ask those students wandering through Gasson Hall today.

Latest Posts
Follow Me

Get new content delivered directly to your inbox.