DDJ: Your background was in mathematics?
JG: My undergraduate was in mathematics and engineering. I received a joint degree. I loved mathematics. It is beautiful. I actually was really interested in the ability of using computers to help us to understand the world around us. That was a thing that was alluring to me.
DDJ: Before we get to databases, can you describe the hardware technology that was available at the time?
JG: This was at a time --1962 or 1963-- when vacuum tubes had been replaced with transistors. That was a big improvement. We still had magnetic core memories, little tiny ferrite cores which were typically made in Hong Kong by people threading three wires through each core.
The machines were executing maybe hundreds of thousands of instructions second. A tenth of a million instructions per second (MIPS). Those were the fast ones. The slower ones, of course, were an order of magnitude slower than that. Then, they had memories that ranged from [about] a kilobyte up to a megabyte. The very largest machines would have a megabyte worth of storage. Punching cards [were] the popular I/O devices.
[To] run your program you would punch your program into these punched cards. That was typically your storage media for your program and your data. You would have a deck of cards typically this long of which this was your program, and this was your data. It was a very physical kind of thing. You would dump this in and it would read the cards in, run your program against the data, and then print out the results. At that time, we were beginning to get disks storing tens of megabytes. If you were lucky, you had a bit of a disk quota. You could keep your programs, and your data, on disks. Then, you would have a much smaller deck with just the delta. The rich people had small decks. The poor people had big decks.
DDJ: Obviously, the development of databases was limited by the available technology.
JG: Yes. Incidentally, what I am describing is the environment at the University of California at Berkeley. I was a poor student. Graduate students were in the same boat. Starting in, I think, about 1958 and, in fact, evolving into the early '60s, there were people in industry who were building on-line databases using disk storage. The Saber system, the airline reservation system, was one of the very first. I think it was Pan Am built the first on-line reservation system. It was a disk space system.
DDJ: What type of data formats were being used? With storage being at such a premium I assume efforts were made to compress the data or to store it in a way that conserves resources.
JG: Here we are and the year 2000 is coming up on us. Indeed, the reason we are having the year 2000 problem is that people were interested in saving two bytes. Indeed, storage was very precious. Record sizes were typically measured in bytes, not in kilobytes. The typical record might be 50-100 bytes and [was laid out] very, very carefully to use the minimum number of bytes that you could get away with. Compression was not used much because, as you know, compression uses processor cycles to compress and decompress. Processors were just as precious as storage in those days. The compression was really going on in the head of the individual. People were packing fields as tightly as they could, but not much beyond that.
