My History in Computing
Communications & Networking – 1966–2016

"Back in the day" of the late 1950's and earlier 1960's, a telephone line and a radio provided connections to the world of news and entertainment.  In and near larger cities, television antennas provided another connection – for part of the day.  Television stations went "off the air" at midnight, replacing the broadcast signal with a test pattern, and many (perhaps most) of the television receivers were black and white.

The world is a bit different now, 50 or 60 years later.

The engineers who created the Internet remember when telephones had rotary dials.  Some of us know the signalling details of dialing a number; most know when Area Codes were first assigned, to allow long-distance calling without talking to an operator.  'DTMF' – Dual-Tone, Multi-Frequency – added choices, reduced dialing time and tempted the first 'Hackers'.

The details all have a back story, with lots of interacting requirements and limits and technical choices, every step of the long path from voice-only telephone lines to the World Wide Web over fiber optics or coaxial cable or wireless, at speeds up to hundreds of megabits per second.

Been there – Did that . . .

1965 Summer - Pelham, New York
Still in high school, Zenith black & white television at home.  Home-brew short-wave radio receiver in the attic, listening on public safety bands and amateur 4M and 2M bands.  Standard black dial telephone, PE 8-4044.
1966 Summer - New York City
IBM 1050 Graduated from high school, working at a summer job in New York City.  Learned to use an IBM 1050 terminal – 134.5 bits per second (bps), poll-response protocol over a multi-drop leased line, connected to an experimental interactive computer system (M44/44X) at IBM Research in Yorktown, New York.
1966–1967 - Freshman Year at MIT
Wandering after hours, learned to operate the IBM punch-card equipment, the IBM 1401 system used to prepare batch jobs for the IBM 7094, and the IBM 2741 and Teletype® terminals connected to the MIT Compatible Time-Sharing System (CTSS).  The IBM System/360 Model 65 running OS/360 used magnetic tapes and punched cards.
IBM 2741 The 2741 terminals for CTSS were connected to the IBM 7094 at 134.5 bps, using a 7-bit IBM line code for the Selectric® mechanism.  The IBM 7750 communications controller and the MIT PDP-1 also supported Teletype® terminals running at 110 bps with 7-bit ASCII (IA5) coding.
On one notable occasion, an MIT instructor called into the CTSS system from England, using a trans-Atlantic voice grade connection.  The Telex network provided very long distance connections at 50 baud, using 5-bit IA2 coding to get 10 characters per second.
1967 Summer - Cambridge, Mass.
Worked as an operating system programmer at the MIT Computation Center, mostly in the machine room or directly connected to the CTSS system.  The IBM ASP/360 software that Fred Abramson and I were working on interconnected two IBM System/360 machines, locally, over a Channel-to-Channel Adapter (CTCA) running at about 1.5 megabytes per second.  The I/O cables were a bundle of shielded coaxial wires, over an inch in diameter.
1967–1968 - Sophomore Year at MIT
Continued my part-time work at the MIT Computation Center doing user support, miscellaneous programming and librarian duties for the SHARE User Group software distributions.
1968 Fall - Junior Year at MIT
Started a part-time position with the IBM Cambridge Scientific Center in Tech Square, across Main Street from MIT.  Interactive systems at IBM were based on CP-67/CMS, using IBM 2741 terminals both locally, via acoustic couplers over voice telephone lines, and via dial-up modems over the IBM tie-line network.
Working with Ed Hendricks, experimented with and developed support for synchronous data communications at 2000-2400 bps, using IBM Binary Synchronous Communications (BSC) between CP-67 and an IBM 1130.
1968–1971 - IBM Cambridge Scientific Center
Left school and continued with IBM as a full-time employee.  Communications over voice-grade lines were typically limited to 2400 bps, though multi-channel and faster modems started to allow up to 9600 bps.  Dedicated lines could operate at speeds up to 56 Kbps using V.35 or proprietary interfaces.
For our early experiments, Fritz Giesin put together a 2400 bps modem eliminator so we could exchange data between the System/360 on the second floor and the IBM 1130 on the fourth.  For one confusing bug, we had to use an analog strip recorder (ink on paper) to verify the actual mark/space signals on the wire.
1971–1972 - IBM CP-67/CMS Development
IBM 2741 terminals were locally connected, but one time I dragged a "portable" terminal (full size Selectric, light blue hard case, built-in acoustic coupler, 30 pounds or so) home to the Putnam Avenue house in Cambridge.  My housemates thought nothing of it, until I sat back and the 'typewriter' starting writing on its own.
1973–1976 - IBM VM/370 Development
3270 Terminal Display terminals – IBM 3270s, 26 lines of 80 chars, mostly – were the instruments of choice.  They were connected to a cluster controller by coaxial cables running at 1.5 Mbps.  The controller was connected to the computer either locally via the byte multiplexer channel or remotely over a 2400 to 9600 bps binary synchronous (BSC) line, running a poll-response protocol.
1976–1978 - Digital Equipment (DEC)
VT52 Terminal The Distributed Systems group was located on ML5-5, the fifth floor of building 5 in the Civil War era woolen mill complex in Maynard, Mass.  The terminals were locally connected at 2400-9600 bps, I think, via 20-ma current loop interfaces.  Current loop was a version of the very old Telex network wiring, useful because it avoided the wire-length limitations of RS-232C without the cost of coaxial cables.
The communications software that I was developing for DEC provided support for the IBM Systems Network Architecture (SNA) protocols, using an SDLC / HDLC interface that operated at up to 19.2 Kbps.  Typical computer-to-computer networking (DECnet) ran over synchronous lines at similar speeds, if I remember right.  Terminal traffic most often was limited to 9600 bps over various interfaces.
1978–1979 - Cambridge Telecommunications (CTX)
CTX was a small company startup with network interface software for the IBM 3704 and 3705 communications controllers, providing terminal support across networks such as X.25, CitiNet, Datapac, Tymnet and Telenet.  Terminal connections ran at 110 to 1200 bps over dial-in lines to a concentrator (PAD - Packet Assembler and Disassembler) - then multiplexed over synchronous leased lines at speeds of 9600 bps to 19.2 Kbps.  CTX developed a remote terminal concentrator for dial-in terminals, based on Zilog Z80 microprocessors.
1979–1985 - GTE Telenet Communications
CTX was purchased by Telenet Communications a year after I joined, which was a few months after Telenet had been acquired by GTE Corp.  GTE Telenet provided X.21 and other terminal protocols over a backbone X.25 virtual circuit network.  Working with other network service providers – Datapac, Memorex, Tymnet, Citibank and larger IBM customers – Telenet developed virtual circuit support for a variety of terminal protocols.  Network backbone speeds were 56K to 1.544M bps (T1) for long-haul links.
PDF Welcome to GTE Telenet Burlington – 1981
1985–1989 - Prime Computer
The 1980's was a period of data communications expansion based on minicomputers, workstations and the new personal computers offered by Apple Computer and IBM.  Networks were running over coaxial cable, over public X.25 services, leased telephone lines, and many other media.  Ethernet and TCP/IP became common, as did AppleTalk and NetWare and PrimeNet. 
JPEG DoD Internet Architecture (TCP/IP and Friends)
1990–1993 - Ungermann-Bass
Ethernet and IBM Token Ring and FDDI were the stars of this period.  Connectivity on engineered wiring supported local network speeds of 4 Mbps, 10 Mbps, 16 Mbps and more.  Speeds up to 100 Mbps were available on early fiber optic networks, especially for trunk lines between network wiring hubs and server locations.  Network hubs and wiring closets were the products, but there was also some early FibreChannel technology at data rates up to 1 Gbps in machine-room locations.
HTML Ungermann-Bass Access/Open (Network Services Platform)
Business use of data communications was spreading wildly.  Ungermann-Bass headquarters and management software development were in Santa Clara, California; network component engineering groups were in Andover, Massachusetts and in Boca Raton, Florida.  There was a three-way video conference system (by Compression Labs, Inc.) setup to allow remote meetings, using a pair of dial-up telephone links at 56K bps each.
1993–1997 - Augment Systems
The need for speed had set in.  Gigabit machine-room networks connecting storage arrays with processors were available over both fiber optics and over engineered wiring.  Simple AppleTalk at 230 Kbps was still around because it was very inexpensive, but 10 Mbps Ethernet was most common.  The Augment AFX-410 servers ran over 1 Gbps FibreChannel Arbitrated Loop (FC-AL) for mixed clusters of Apple Macintosh and Windows-NT PCs.

Integrated Services Digital Networking (ISDN) provided higher data rates and new services, based on technology updates in the 'telco' networks.  The Internet and the World Wide Web became available to the general public at useful speeds.  Specialized processors and software dropped the cost of connections dramatically.

1997–1998 - VideoServer Connections
Dialed a live video call at 384 Kbps, from our lab in Marlborough, Massachusetts to the Teletek lab in Taiwan, via three ISDN Basic Rate interfaces and a prototype end station that cost less than $7000.
1999–2000 - NorthStar Internetworking
Dial-in terminal server for Internet access: up to 24 channels at 56 Kbps over T1 interfaces (U.S.), up to 31 channels over E1 interfaces (CCITT).
2000 - Hammer Technologies
Multi-channel, multi-protocol data network behavior testing and network simulation at interface rates up to 2.5 Gbps, using custom processors (C-Port C5) in a Compact PCI chassis.
2000–2003 - Vivdon & StarBAK
Multimedia over the Internet, music streaming and video conferencing, based on custom operating system software and soon-to-be-generic Intel server engines.
2003–2004 - Katana Technology (Virtual Iron)
Distributed virtual multiprocessor with shared memory, in a server cluster exploiting Infiniband at 12 Gbps.
2004–2010 - Crossbeam
High performance network traffic with security policy applications, in redundant network switches handling up to 40 Gbps aggregate data rates.
2011 - SafeNet
High security network file system (NFS) servers with encryption, using Freescale multicore processors.
2011–2012 - Symmetric Computing
Distributed multiprocessor Linux clusters with very large memory, direct-connect Infiniband at up to 40 Gbps.
2012 - VideoIQ
Analog and digital video encoding with real-time activity analysis for surveillance, 10/100 Mbps Ethernet with Internet access.
2013–2015 - Affirmed Networks
Subscriber Services and Content Gateway for mobile network providers, using clustered servers with multiple interfaces at 1-10 Gbps.
2015–2016 - Avigilon
(New corporate parent for VideoIQ technology.) Real-time video analysis and network-edge recording, supporting 10-100 Mbps Ethernet camera network connections and 1 Gbps backbone networks.

Copyright 2017 by David B. Tuttle, Reading MA
This page last updated 19-Mar-2016