Wednesday, May 15, 2013

My Life With Technology - Chapter 2.0

Detour to Dalton 

by Bill Holmes

Around 1971 I had a slight career change. I had been contemplating leaving First National Bank of Atlanta (FNB) for a variety of reasons. I had at least five different positions in about two years and had become something of a Mr. Fix-it. While it was flattering to have the bosses' confidence it was also exhausting. I was working tons of hours and all kinds of crazy shifts. It wasn't hard for me to get used to a straight second or third shift but it was hard to be constantly changing work hours. When you work at night there are still work related activities and meetings that require you to show up during the day. The bosses never schedule a staff meeting for 2:00 AM. Another problem is that the day shift people have no compunction about calling you with a question in the middle of the day while you are sleeping. Somehow that was OK but if I called them at 3:00 AM it wasn't. At one point the second and third production control shifts were having trouble communicating and handing off work. I was put on an 8:00 PM to 4:00 AM shift to ensure continuity. That is the worst shift in the world. Too early to have evening activities and too late for any after work activities. Luckily they reorganized the department soon after that from a functional to a shift management structure and I took over third shift. There were a few other reasons for my dissatisfaction. I don't remember if I actually turned in my resignation or just told them I was planning to leave.

Apparently word got around that I wanted to leave and I got a call from a manager I barely knew. He was starting up a new department that was going to open and staff processing centers around the state. His offer was that I would help set up and open the first remote center in Dalton, Georgia and then set up, open and manage the second center in Valdosta, Georgia. There was a slight pay raise involved too. It sounded exciting and the fact that I'd soon be about 250 miles from the closest boss was a bonus. Well, it was exciting but starting a new department and opening data centers didn't reduce my workload in fact it increased.

The data centers were to process checks and print reports for smaller banks around the state. The new centers would let us sell our processing services to banks farther away from Atlanta. This was similar to the services we were already providing but in a more distributed and far reaching manner. We were also able to eliminate many of the paper tape transmissions into Atlanta from distant banks. These remote data centers were basically input and output locations. The actual account posting was still done on the big computers in Atlanta. Each center would have a small computer (power, not size), check reader/sorter, card reader and printer.

This initial center was housed in the First National Bank of Dalton about 90 miles north of Atlanta. Dalton had a population of around 20,000 and the county 60,000 at the time. A little smaller than Atlanta. Dalton was and still is the carpet manufacturing capital of the world. FNB Dalton was a fairly big bank because of all the carpet mills. This was going to be a hybrid operation. FNB Atlanta would set up the center, train the personnel, lease the equipment and provide technical support. Kind of like a franchise. Day to day operations would be staffed and run by FNB Dalton people once they got trained and the initial kinks were worked out. I think the two banks had some other business relationships and in fact FNB Atlanta bought the Dalton bank several years later. FNB Dalton had just built a new four story building and they had included a small computer room with raised floor in the design. They also had a freight size elevator so physical planning and equipment installation was relatively easy. Now this computer room was a little unusual. One wall, the biggest, of the room was covered in red/white/black long shag carpet. It actually helped deaden the noise but it was a dust magnet. Processing checks, cards and running a high speed printer generates a lot of paper dust. The janitors had to vacuum the wall every so often. The whole bank building was a showroom for carpeting including what was at the time the largest carpet sculpture in the world. It was on the wall behind the tellers in the two story main banking floor.

But you didn't come here for the history of carpets so on with the more technical details. By the time I got involved, the equipment for the remote data centers had been decided on. Although Burroughs equipment was used for check processing in Atlanta it was decided to use IBM equipment in the field. There were a couple of good reasons for this. The remote centers would be communicating with an IBM mainframe back at FNB and it was much easier to get IBM support out in the hinterlands. Also these new data centers were under the management of the FNB Data Processing Department which had all IBM equipment.

The equipment configuration at the remote centers was an IBM 360/20 computer, IBM 1419 MICR reader/sorter, IBM 1403 printer, IBM 2501 card reader and an IBM 029 keypunch. In addition there was a leased data telephone line and a modem that was about the size of a toaster oven.

GTE  Phone prior to data phone mods
This was 1971 and data transmissions were pretty primitive. A leased data line was a point to point connection provided by the telephone company. They were actually very similar to normal telephone lines but with a little extra conditioning. It was not one continuous cable from point A to point B. These lines were of course analog, no digital internet or fiber optics. Because it was an analog network, it required modems to convert the data from digital to analog and back to digital. We used 4800 baud modems, blazing fast for the times. A baud is approximately equivalent to bits per second (bps) but there are many variables. It's not a straight forward conversion. Lets just say it was slower than your current home broadband internet connection and probably your cell phone. You could also talk over a leased line although you were limited to the one phone on the other end. Data line phones looked like a standard phone without the rotary dial and with a small knob/ button to switch between voice and data modes. You turned the knob to switch modes and you pushed it down to ring the phone on the other end. They were usually grey. We had an extra challenge in Dalton. It was a General Telephone (GTE) franchise area. GTE was not quite as up to date as the Bell Telephone franchises. Ironically I worked for GTE in a later life. Near as I can remember there was only one telephone installer in Dalton who knew how to do data lines. He had to take a regular rotary phone and modify it for a data line. The modification included drilling a hole through the outer plastic case right of the 0 (Operator) to install the data/voice switch, a real confidence builder. The other problem was that as mentioned Dalton had GTE local phone service but AT&T controlled the long distance lines and Southern Bell (an AT&T RBOC) the local service in Atlanta. As we all know, when two or more vendors are involved if there are problems there is often finger pointing and "everything on my end is OK, must be the other guy's problem". After many tries over several days the data line was installed and finally working.

I mentioned the toaster oven sized modem. This thing had dials and switches and lights and a VU type meter that measured signal strength. It was also manually adjusted.  Modern modems are self adjusting, balancing and equalizing but not this big guy. When it got out of wack you'd have to get on the line with the other end and play with the dials to get it back in sync. Much like tuning a radio to a far away station. The modems had a test mode which generated a signal loop between the two ends of the circuit. The meter indicated signal quality. Sometimes the line quality was especially crappy and it was almost impossible to get the modems to talk to each other at an acceptable level. By paying for a leased line, we had access to the guys who monitored the telephone networks 24/7. We referred to them as the long lines test board. You would give them your circuit number and they could run tests end to end. Of course you couldn't use the line while they were
checking it out. If there was a problem, they could usually isolate it to a segment of the long line network or the last mile of the local telco. If they determined a segment of the wired network, as opposed to microwave segments, was the problem they could burn it. That meant sending a blast of electricity through the wires which would heat them up and evaporate any moisture. That usually fixed the problem. The guys at long lines were pretty good. They mostly dealt with internal telco people so they were more candid. Since there were so few data lines we got to know each other pretty quickly and were on a first name basis. They would actually admit when there was a telco problem and then try to fix it. If they couldn't fix it from the test board they would dispatch someone or tell us to call the local telco. They's give us a trouble ticket number that we could pass on to the local guys which helped. We got pretty good service, of course we paid a whole lot more for our telephone line than a normal customer.

IBM 360/20

Our computer was an IBM 360/20. This was the smallest and cheapest computer in the 360 family. I think the ones we used had 8K of memory. The biggest 360/20's at the time only had a 16K memory. It was primarily designed to replace some of the old tabulating equipment. We used it to drive the check sorter, printer, card reader and handle the data transmissions. The Model 20 had no typewriter console. The operator communicated with it via dials, buttons and lights. It also had no storage device(s) so when we loaded checks on the sorter they were transmitted real time to Atlanta. The same when we were printing reports sent from Atlanta. That caused problems that I'll document in following chapters.

IBM 1419 MICR Reader/Sorter

The IBM 1419 MICR reader/sorter was probably overkill for this installation. It was IBM's premier sorter at the time and processed 1,600 documents per minute (DPM). It had 13 pockets, 0 through 9, A, B and reject, just like a card sorter. It was connected to the 360/20 for program control and data transmission. Like the Burroughs sorters, it could do numerical sorts offline from the computer.

IBM 2501 Card Reader

The 2501 card reader was the cheapest card reader available from IBM and read 600 cards per minute (CPM). It was used to load programs and read the cards punched from the sorter rejects.

IBM 1403 Printer

The IBM 1403 printer was an old but still viable device first used on the 1400 series computers that predated the 360 family. It was a chain printer that ran at 600 lines per minute (LPM). That was fast enough to keep up with the sorter and the transmission speeds.

I'll get into more specifics in the next chapters, but basically we would run the checks and deposits through the sorter and simultaneously transmit the data to the mainframes at FNB Atlanta and print a listing of the transactions. We would balance the run, keypunch the rejects and make any corrections/adjustments. We would then transmit the punched cards to Atlanta. Once everything was in balance we would give Atlanta the OK to post/process the bank. They would do that processing on their schedule. While waiting for that, we would begin to sort the items into account number order. Once the bank's checking accounts were processed, Atlanta would transmit the reports back to our printer. It was all pretty amazing for the times and also rife with problems.

Since this was a new venture for FNB we were on a rather limited budget. This was a little skunk works project by the Data Processing Department, not a big deal like the BankAmericard venture. There were two of us in Dalton on and off to set up the data center and then five days a week for a couple of months once we got started. During the set up we mostly drove back and forth from Atlanta and didn't stay overnight. It was a 90 mile trip one way via Interstate 75, a very good road for the times. We drove a little over the speed limit so it wasn't a long trip. Once we started live processing we needed to be there all the time. The other guy I was working with had relatives in Carbondale, GA, about 10 miles south of Dalton. The relatives worked in a carpet mill. We stayed in their spare bedroom rather than a motel. That wouldn't have been so bad except the spare bedroom had a double bed. They were all nice people but it was not the ideal situation. I much prefer sharing a bed with a female. FNB paid them something for our lodging. We would drive up to Dalton on Monday and stay until Friday night or Saturday morning then drive back to Atlanta. The following Monday we'd do it over again. After the first couple of weeks we did switch off on the Friday late shift so one of us could get back to Atlanta by late evening.

The first couple of weeks after going live were a cluster f#*k. I was the only one who understood the check processing procedures and could balance the runs, the keypunch operator was new to the machine, none of us were familiar with the 360/20, the data line was spotty at best, there were some programming errors that didn't show up during testing, the local IBM guys weren't familiar with the maintenance of all the equipment and the guys in the Atlanta computer room were learning remote processing too. Things happened that we never anticipated. There were a few supplies and pieces of equipment we forgot to include in the initial setup. We managed though and since we started with just one bank we had some slack in the schedule. I can not remember whether the Dalton data center ever processed other banks. I had some technical support dealings with them over the years but since it was staffed by FNB Dalton people I had far less contact with them than the other centers once I left.

Once we got up and running it was mostly a second shift operation. We couldn't get hold of the checks to process before 3:00 or 4:00 PM, later many days, and couldn't finish printing reports much before midnight. There wasn't much to do in Dalton at midnight. I think the town and/or the county was dry except for "private" clubs. Being foreigners we didn't belong to any of those clubs. We'd just find a greasy spoon, have a meal and then go to our luxury accommodations.

I knew this was a short term assignment and I would soon be off to start my own data center. That made the commuting and less than ideal lodging tolerable. I learned a lot which helped me avoid many of the same pitfalls. Of course being very resourceful, I managed to find a whole new batch of pitfalls and problems on my next stop.

That next step was to go back to Atlanta for a few weeks to work on the preliminary planning for the Valdosta center. I went back to Dalton a few times to help out but I didn't stay overnight. I made a couple of trips to Valdosta too during that time.

I'll go into more detail about our remote data centers in subsequent chapters. Pack your bags, we're getting ready to permanently move to the other end of Georgia.


No comments:

Post a Comment