Sperry Univac [1973-77]

My employer wanted to groom me for a career in life insurance management. My heart was in Data Processing. After three years, I left to join Sperry Univac. My goal was to spend more time programming. Unfortunately, the machines were so expensive, we could not afford a unit in our little branch, so I was forced to beg and borrow time on customer machines. The only time I got near the machines was to solve a problem that stymied the customer's programmers. Since they spent more time on the machines than I did, they had more experience, and I was put in the difficult position of knowing less than they did.

My first training course with Univac was an Assembler class, in New York (1973). During my spare time I visited all the sights, including the World Trade Centre. I was always a wise guy trying to pull stunts. When we learned how to submit jobs to the 9400, I discovered one job could start another. So naturally I quickly prepared two jobs, A and B. Job A would start Job B, which in turn start launch Job A. Once they were launched, they tied up the system and could not be stopped without rebooting. Just as my prank unfolded, the manager of the training centre visited the computer room and asked my panicked instructor how things were going.

I worked in Montreal, then Calgary, where the oil business created many opportunities.

The branch was divided into three dominions: the salesmen, the Systems Analysts (SA) who helped with demos and benchmarks, and the Customer Engineers (CE), who maintained and repaired the equipment.

The benchmark was the center of the sales cycle. It consisted in converting a customer's application from his existing equipment, to a Univac model, and demonstrating how much faster it could run. I've had good benchmarks, and I've had a few failures.

I recall in particular the ATCO benchmark of December 1977. ATCO was a big manufacturer of trailers, used as offices or dormitories in construction sites and other places requiring temporary buildings or accommodations. Their trademark white and yellow units can be seen all over the world. ATCO wanted to upgrade from a DEC system. We pitched an 90/80, the top of the line of the Series 90.

We flew to headquarters in Blue Bell, near Philadelphia. For weeks, we struggled to execute a massive conversion, involving a data base as well as a transaction processing system. Working around the clock, we got depressed and homesick. During our brief sorties for food, I remember hearing the old tune "I'll be home for Christmas", and feeling nostalgic for my family. We were not able to complete the conversion in time. The customer flew down on Univac's Lear Jet, with wall-to-wall liquor. We flew back to Calgary on the jet. No liquor.

The VS/9 operating system team had not completed testing the drivers for the latest model disk drives. We started getting irrecoverable I/O errors during the data base tests. That was the beginning of the end. IBM won the benchmark and got the order. Ironically, ATCO suffered huge cut-backs a few months later, and cancelled the order.

On another occasion, I was sent to Blue Bell to take part in another benchmark involving 9300s and the flagship 1100s. We were relegated to the graveyard shift. One of our programmers was a draft dodger, so he had to stay in Ottawa to work on the benchmark. One night, the Blue Bell team was arguing with the Ottawa team that they were sending the wrong codes over the line. It was not until several hours later that it dawned on someone that one team was talking hexadecimal while the other was using octal. The octal team was asking for ‘40', which was 000 100 000 = X'20', while the hex team was sending X'40' = 0100 0000 = octal 100.

We made lavish use of our expense account, arriving one night very inebriated after imbibing copious amounts of ale. One guy turns around and says, "You know, we often talk of pissing on Univac, but we're never done anything about it. Well I've got an idea". So the three of us lined up against the wall, under the huge Univac sign glowing in the night, and relieved ourselves in the heart of the realm.

Univac was a pretty good place to work. The salesmen fancied themselves the cowboys that wouldn't fit in at straight-laced IBM. It has been said that every good salesman has a theatrical penchant. Univac had its share of imaginative, daring pitchmen who knew how to cultivate customers. We once took a prospective customer to dinner. As the waiter poured a flaming Irish coffee, the salesman turned to the customer and says: "Every morning at IPL, you get this!" (IPL is Initial Program Load, when the computer is bootstrapped).

The Halifax accounts were sold by pimping and pandering as much as anything else. DP managers were regularly brought to Montreal for hookers and pot.

CADE

I owe a great deal of travel to a little machine called the CADE (Computer Assisted Data Entry). The CADE 1900 was manufactured by Pertec, and re-sold by Univac. It consisted of a mini-computer, connected via coax to 32 workstations, where each operator could pump up to 12,000 keystrokes/hour to a 50 MB disk. The data was transferred to tape, to be processed on a mainframe. The unit also had RJE (Remote Job Entry) capability.

My first experience was when we proposed a CADE to the Prince Edward Island government. The only available unit was in New York City, so I got shipped off to prepare a demo. The salesman and customer followed a few days later. This guy was a lush, so we hit every tavern up and down Times Square, and wound up in a theatre showing The Devil in Miss Jones. A real trench-coat establishment. We sneaked a bottle of liquor into the theatre, but we didn't have any drinking glasses. So we march up to a vending machine, punch a selection, and deftly whip away the paper cup as the ice and soda go cascading down the drain. A wino shakes his finger and sputters: "But, but... you're losing the drink!"

The first time I saw Salt Like City was on a CADE benchmark.

The CADE was exclusively a data entry machine. However, because of its small size, it was the closest thing Univac had to a mini-computer. Some misguided salesmen insisted on pitching it as a full-fledged computer. They just couldn't understand that its rigid, data-entry-oriented design made it unsuitable to compete with DEC's PDP series.

The Alberta & Southern Gas Co. (know as ASGAS) pumped natural gas through their pipeline system, for a variety of customers. The quantity of gas was calculated based on pressure and volume, which were recorded by pens drawing on circular charts. The area under the charts was computed by special optical scanner/integrators, which printed the result on paper tapes. These tapes were then manually re-keyed by keypunch operators, a time-consuming and error-prone process. ASGAS wanted to know if we could automate this.

There was no existing technology to connect the scanners. However, we located a bunch of young engineers operating out of their garage, showed them the confidential schematics of the CADE keystations, and within a few weeks they had a prototype circuit board that received the pulses from the scanner, simulated the keystrokes of a CADE keyboard, and transmitted the data to the CADE. This one-of-a-kind hookup cost $10,000, and netted Univac a $100,000 sale.

The CADE is what got me to Calgary, and introduced me to the wonderful and wacky staff from that office, as well as the gorgeous Canadian Rockies. I transferred there for three exciting years.

It's probably the only time I was upbraided for submitting expense claims that were too low - my boss told me it would make the other guys look bad.

Quota trips

My boss, a great guy by the name of Bill Meyer, had a custom of rewarding the SAs with their own "quota trip". The salesmen who achieved their quota got to go to exotic places like Rome, Paris or Hawaii. Not to be undone, Bill would haul us off to Kalispell, Montana, or Revelstoke, BC. We would pile into the CE manager's RV, loaded with beer, and drink all the way to our destination. The CE honcho, an old scot who loved to sport cowboy boots and a bolo tie, did the driving and, of course, did not drink until we got to destination.

I remember downing way too many tequilas in Kalispell, in a bar with swinging doors and sawdust on the floor. As I staggered out in the wide, western street, my last conscious memory was being sprawled on the sidewalk, my collar caked with vomit, staring up a pant leg with a characteristic yellow stripe down the side. How I got to my motel I will never know, but I am grateful to the patient and tolerant Kalispell constabulary. That, by the way, is the only time I ever got sick from drinking.

The Calgary SAs were a tough bunch. On one trip, they went through the Rockies on the legendary Canadian National Railway dome car. They got really plastered, and their rowdy behaviour annoyed the conductor. When they reached their destination, a couple of guys forgot to get off. When the train started to leave the station, they got up in a panic, but the conductor (a hunchback they called ‘Quasimodo') refused to stop the train. So one of the SAs points to the glass dome and says "If you don't let us off, we're going to barf all over the inside of the dome, and you wouldn't want that, WOULD YOU?".

The conductor stopped the train right there, and let them off. They walked back to town along the train track.

One SA, with more courage than judgement, tossed a loose-leaf manual out the window of the eight floor of a hotel room. Downtown Calgary was showered with with a cascade of white confetti, labeled ‘Univac'. He was fired the next day.

Our branch manager, Don Featherstonhaugh, is reputed to have had some bizarre experiences on his quota trips. In Paris, he got a haircut that looked like the proverbial bowl of soup had been placed on his head. Coming back from Rome, the passenger next to him died of a heart attack, and he had to spend the rest of the trip next to a cadaver.

Univac had a CRT terminal called the Uniscope. Some models included a cassette tape to record data. The unit could display a form or mask, and operate off-line. There was a rudimentary, sequential search capability, to allow for editing and correcting of data. One guy, who transferred from accounting to sales, wanted to sell it as a random-access device, using the search facility to retrieve records based on a key!

Prior to acquiring the RCA business, Univac was struggling with an operating system called OS/7, which was to run on the 9700, the successor to the 9400. Prior to the release date, I heard rumours that the OS team was still having problems with the ‘getmain' function, which provided the caller with a chunk of memory, and was crucial to the operation of the OS. Such major instability no doubt made it easy to axe OS/7 in favour of RCA's VMOS (Virtual Memory OS), later renamed VS/9.

When Univac announced they were ditching OS/7, a disgruntled colleague, Bob Milliken, who had spent considerable time learning OS/7, circulated a mock advertisement of a toilet bowl with the caption "OS/7 - liquidation sale".

We benchmarked a 90/30 for Calgary Coop, proposing Uniscope CRTs for on-line credit validation. After spending a few weeks preparing the demo, we invited the Controller and his DP manager to Mississauga, Ontario. I'll always remember the smile on the dour Controller's face when I keyed in an account number, and on the screen appeared the account details. I knew then we had the sale.

I was also briefly involved in the City of Montreal account. The City was a major 1100 customer, with satellite 9300s used for print spooling. We were brought in one day because employees were abusing the computer by deliberately hammering on switches at random, trying to provoke a job failure and re-runs, in order to get overtime work. We were not able to re-produce and fix the problem. Computer systems are complex and difficult enough to make right, without the added poison of saboteurs screwing things up.