How to build computers for big data, 1950s style

IBM has a new security computer cleverly mounted in an 18-wheeler. There is precedent for this going back to the 1950s and beyond.

Image: US Army report

IBM recently installed its new Security Operations Center in a tractor trailer, which is a smart decision for marketing and field use alike, but it's hardly a modern innovation: IBM and the US government were doing this seven decades ago.

The idea wasn't new even back then. Horse-drawn carts moved telegraph equipment in the Civil War, while a combination of mules and trucks carried radio gear during World War I. IBM and the US Army developed Machine Records Units, which were pairs of 18-wheelers loaded with punch-card accounting hardware that followed every Allied corps worldwide.

Then, in the first half of the 1950s, the computer industry became less experimental and more commercial. The US Department of Commerce tasked its National Bureau of Standards (today called the National Institute of Standards and Technology) with being a clearinghouse for government computer consulting, purchases, and pure research. NBS staff did research on memory, input-output methods, signals timing, and logic circuits.

One of the NBS testbeds was a computer called DYSEAC. "DY" meant second; "SEAC" was the Standards Eastern Automatic Computer. But where SEAC in 1950 lived in a stable room, the team behind DYSEAC designed it to fit into two 40-foot trailers. It would be a mobile computer, moving as needed just like the telegraph carts and accounting trucks before it.

NBS produced a single DYSEAC in 1953. Five years later, its designers wrote:

"In the course of the planning for certain types of control system experiments, it became evident that considerable advantage would accrue if DYSEAC could be moved to the actual operating site. The packaged modular form of the machine... readily lends itself to a compact and sturdy construction suitable for transportable service. Accordingly, the program was altered and appropriate structural modifications were incorporated so that the machine could be installed in a trailer van... Thus, DYSEAC could operate as an independent installation in an isolated field location."

Engineers beefed up everything in DYSEAC from its mercury acoustic memory in Pyrex glass tubing, to its modular components for quick replacement in the field--some of which used printed circuit boards--quite possibly the first time this ever happened in a computer.

"It is believed that the use of printed-circuit techniques will materially reduce the troubles that have resulted from the wiring methods that have been used in SEAC. In addition, the need for more rugged construction of removable components which has been indicated by SEAC experience has been incorporated into the DYSEAC design," the engineers wrote.

The computer had 1,024 I/O units and 4 kilobytes of 48-bit words. NBS delivered it to the White Sands Signal Agency in 1954. A test application was the automated evaluation of wind forecasts for White Sands' Missile Geophysics Program. "The manual reduction and analysis of this data involves a large number of man hours," technicians noted in a report. "Use of the DYSEAC computer in reducing winds aloft data is highly feasible for the following reasons: (1) It reduces computational time; (2) It eliminates the possibility of human errors resulting from hand calculations; (3) It provides an easy check for any errors made in feeding the values into the machine; and (4) It requires fewer personnel to produce the end results."

DYSEAC's rugged design and testing were impressive for its day. Unfortunately, its testing in New Mexico exposed an Achilles' heel. "It was an unfortunate destination because the white sand at this proving ground infiltrated the computer circuitry, causing it to malfunction," head engineer Alan Leiner wrote. "But the logical design of this machine was a significant advance towards our ultimate goal of organizing a network of computers to meet deadlines."

(Author's note: This article is adapted from Chapter 1 of my book, Abacus to smartphone: The evolution of mobile and portable computing.)

Also see