Being first to market, ARCnet

I’m going to date myself. How many of you out there remember ARCnet? In 1976, I was attending Drexel University. As an early geek, I lived in the computer center on the basement floor of the student center rarely attending class. My days were filled experimenting with everything from dumb terminals connected to early mainframes, minis and micro computers to teletype systems with paper tape or punch cards. Drexel provided me such a wonderful exposure to technologies and early devices such as the IBM 5120 and the Textronics Plotters and Datapoint Databus CPU’s to name a few. Drexel also had a mainframe Burroughs 5500 and RJE Terminal to UNI-COLL, a shared non-profit hosting an IBM360 at first and then an IBM370 mainframe for some thirty higher education institutions on 34th and Market Street in Philadelphia. It was part of the Science Center back in the day of timesharing. I will cover some stories on UNI-COLL another day. This day, I will reflect on ARCnet.

As a student, I got into everything. Nick Demaio, the Drexel Computer Center IT manager grew fond of my skills and one day gave me a job of looking into the Datapoint Databus CPU employed in the Admissions department. I recall a group of CPU’s were connected together and they looked like hybrid office machines – part printer – part terminal – part CPU. They were used by administrative operators in Admissions to follow-up inquiries and applicants. In other words, they were early word processors, designed to support the text letter management and data files storing all the names and addresses of people who contacted the University. The applications were developed by Michael McCabe, who was a few years ahead of me. The Cobol like language was used with a compiler and the routines developed split across the multiple CPU’s. Multiple CPU’s? Everything that I was exposed to prior centered around a single CPU or a system built with multiple CPU’s but managed by a shared operating system like MVS on the mainframe.

What intrigued me was the connection between the “micro” computers. It was a coax cable on the outside connected to the back of the chassis far more advanced than the punched card readers and paper tape readers we had in the computer center. The applications were developed and managed with no release management, check in and check out or any thought about architecture. The programming to support the Datapoint applications also had to incorporate updates of the OS (operating system) and device driver updates for the plug and play expansion boards. Each board included a set of disks. And, the system software had to be loaded in ROM and made part of the boot sequence. Part of the responsibilities of the computer center was to provide ongoing maintenance of the Datapoint system, as new releases were received on 8 inch floppy disks from Datapoint.

That was where I came in. Nick Demaio asked me to update the system one day. That is an over simplification. I had to learn how to install a set of patches and to re-compile the Cobol programs Mike McCabe developed. There is an art to developing new solutions. Just like there is an art in trying to figure out things someone else created. In the process, I consumed the manuals and technical release papers on the ARCnet and began dreaming of all sorts of applications that could connect computer to computer together. There was little documentation on the actual application Drexel used. Maybe twenty or so pages written into ASCII text files. But, I figured it out.

Reverse engineering was a term that was coined years after I was working on these Datapoint systems. Back then, the data files were localized on one CPU as it served others. Multiuser support had to be handled with file and record locking. ARCnet introduced new challenges, in that use of the tables was no longer under the session control of a single CPU. On top of that, ARCnet had no central CPU. The hub of computers shared the external backbone over the 1.5Mhz communication channel offered by the coax cable. Datapoint’s incarnation of ARCnet was more like peer to peer networking. Do you recall NetBios? ARCnet was the forerunner of that which Microsoft fashioned a decade later and pushed for a long time. We still have NetBios installed over TCP/IP on most networks today. Some of you may be familiar with that. Most of you don’t recall ARCnet because it was the first real venture into (LANS) local area networking before Novel picked it up and supported it as a topology in the late 70’s. It was before Banyan Vines. And, it was before Ethernet and Token Ring came on the scene in the early 80’s as well as a competitive threat.

Token Ring from IBM took the same path as ARCnet. I worked on that too. IBM controlled the technology and did not want other vendors to tell them what to do, so I would call it a pretty closed system. As a result, Ethernet, which was pushed by Xerox and Digital, took off, because it used twisted copper and was open to all to support from Apple to Zenith. What limited ARCnet, Token Ring and other network topologies of that day? Most likely, the reason none of us use them today, was the reluctance of Datapoint to support anything other than RG62 Coax. The engineers thought it was the best way to send signals without interference. From a business perspective, they also thought they had the market cornered and after going public in the early 80’s with tons of capital, they faltered when their senior management and the SEC clashed over revenue recognition, throwing the company in a tail spin they never recovered from.

Meanwhile, IBM and others who resisted open standards and market forces believed Token Ring was their best way to connect computers and devices even with their bulky cables. All their hardware offered Token Ring adapters and none of it connected with Ethernet or ARCnet. That is another story.

The lesson learned from ARCnet and IBM, at least from a marketing perspective, is that sticking with your guns and trying to control the evolution of technology is often better served when one adopts an open market perspective. By opening the Ethernet technology as platform commerce could expand on, we all benefited and we now have the Internet, home computers and a world connected.

If you want to read more about ARCnet, check outhttp://en.wikipedia.org/wiki/ARCNET