DDJ: You mentioned "SQL". How did that, very quickly, come about?
JG: Those other guys (we were the plumbers) were trying to make relational systems easy to use, develop the language, which was originally pronounced S-E-Q-U-E-L because it was a SQL to earlier languages. That name was trademarked so they took the vowels out of it and it turned into SQL. The IBM relational group built alltheir stuff. We built transactions into SQL. We built non-procedural access. We built security. We built, of course, data definition and data manipulation into SQL. SQL was the IBM language. There was a database task group for relational that was now competing with the DBTG task group. IBM brought SQL to that task group and said, "Here, take the SQL language." This happened in 1970. There was this ISO committee that now went ahead and made SQL as standard language.
DDJ: So that is when these other players were able to take language...
JG: They actually predated that. They actually read the research reports and said, "You know, we need a language. Here's a language. We are programmers. Let's get with it. This is as good as any language." Incidentally, the folks at Berkeley had their own language. It was called Quill. To this day you can have debates between the Quill and SQL guys about which language is better. Both of them are better.
DDJ: Let's go back to the point where Software EG comes in the picture and IBM feels the pressure.
JG: IBM had the group of people in Endicott, New York who were chartered with building an easy to use database system. They came to research and said, "What do you have?" We said, "Take this tape." So, in about 1982, is that right, SQL DS came out which was the low end SQL on IBM DOS machines. Maybe it was 1981. DOS was still the mini main frame or not the MS DOS that most of us think of today. That product came out and was quite successful. In fact, stemmed the Software EG Adabase tied, to some extend, in Europe. This created a problem. Incidentally, in that era, then, Informix, Oracle, and... No. Sybase was not on the scene yet. You had IBM product, Oracle, and Informix.
None of these products shared any code base. For example, now it is common for some Unix application. A lot of those were actually developed in academia through a free software open source.
JG: Ingres actually split. Ingres had the Ingres product and then there was University Ingres which continued to be used in the academic setting. They got independent existences at that point. Oracle and Informix did not. They started from scratch. The mainframe guys felt like they were disadvantaged.
A project that I had worked on at IBM eventually turned into DB2 for the high end. In the meantime, I went off and started working at Tandem Computers. Tandem Computers also, at that time, had lots and lots of very talented people. Among them, Jerry Held, who was the Project Lead for the Ingres project at Berkeley. Over time, we ended up building a parallel SQL system at Tandem, which was called "Non Stop SQL" which is now called SQL MX. This was a highly paralleled distributed SQL database. I guess 1988 is the first year that the fully parallel system came out. SQL systems came out and it is difficult to remember that as late as 1987, none of the SQL databases that were products -- The IBM product. . . I guess I have to be fair. None of the Unix SQL products or. . . used more than one disk. They were all mostly for decision support,mostly ad hoc, mostly very simple. I am sort of giving you a history of the SQL products. In about 1981, Bob Epstein had left the Ingres project and was working at Britain Lee. Britain Lee was making a database machine. Epstein decided he wanted to do his own thing. He had some ideas of his own. He went off and started a new company which was called "Sybase". This was very late in the product evolution to come on the scene, but Epstein had an idea. He had the idea that you could have stored procedures. His idea was clients over computing. So you would have clients which would send requests to servers and the server would execute a stored procedure.
DDJ: This was a new idea?
JG: Well, there are no new ideas. However, none of the other SQL systems had stored procedures.
DDJ: But it was in the air?
JG: Well, it wasn't really in the air because nobody thought it was very important. I thought it was important. Bob thought it was important. There were a lot of people who did not think it was important because nobody else did it. When he did this all the sudden he had a factor of three performance advantage over everybody else. Everybody else was going back and forth and back and forth between the client and the server. Here is Bob with a 3X - More, or less, a 5X, less network traffic, and 3X, less performance. All this hemming and hawing about TCPIP or whatever protocol you are using. . .
DDJ: You would just maintain the stored procedure on the server because it is just easier?
JG: Lots of people.
