Bell, Chamberlin remember Ted Codd

The father of the relational database, Ted Codd has died aged 79.

With his background in mathematics, the British-born researcher laid the foundations for a breakthrough in database technology at IBM.

"Ted had the key insights and developed the mathematical theory on which RDMBS were based," Don Chamberlin told us today. "System/R were the carpenters who came along and implemented the ideas."

But while the relational database was lauded as a theoretical concept, the carpenters had a job to convince a skeptical world that it could be a practical reality.

System/R was the code name for the legendary IBM research database project based in San Jose. The work culminated in IBM's DB2 database and included large contributions from Jim Gray, now at Microsoft Research.

"At first it wasn't clear that you could build an optimizer that would be as efficient as a human programmer," says Chamberlin.

"It took a little while for the relational ideas to catch on. It required a software layer to be implemented that could take a high level language that could map it down to an efficient execution plan," he says.

"We had the same arguments with high level languages with Fortran in the Sixties. There would be programmers who would insist that 'I can do a better job with my registers'".

"It took a practical demonstration that we could build a good optimizing compiler for SQL," he said. "We were up against existing products, some of which from IBM."

Mike Ross, a spokesman for IBM's Almaden research lab, sets out why:

"The Year 2000 issue was the result of memory limitations, but we can't imagine not having four digits on a date now; and relational databases allowed lots of flexibility when people weren't used to having flexibility."

Codd is lauded for defining the twelve rules of a relational database.

Although Codd was given an IBM Fellow award and focused on natural language processing, Codd also contributed to the System/R team.

"There was a good synergy. Ted had studied Mathematics at Oxford, other members had more of a programming language and system implementation background."

Even before his landmark "rules", Codd's influence was being felt beyond IBM.

"I was at Carnegie Mellon University when he was the IBM liaison at CMU," Gordon Bell told us today. "It was while I was talking to him that I came up with an invention." The invention was Unibus, an interconnect that Bell introduced into Digital Equipment Corporaton's PDP line of computers. Bell described it as an "aha" moment. He went onto lead engineering on the VAX, and [oral history]

Back to San Jose. Chamberlain related how he Codd spoke in aphorisms, which is a nice detail:

"Codd had a bunch of ...fairly complicated queries," Chamberlin said.  "And since I'd been studying CODASYL <the language used to query navigational databases> , I could imagine how those queries would have been represented in CODASYL by programs that were five pages long that would navigate through this labyrinth of pointers and stuff. Codd would sort of write them down as one-liners. ... (T)hey weren't complicated at all. I said, 'Wow.' This was kind of a conversion experience for me, that I understood what the relational thing was about after that." [via the fascinating and exhaustive System/R reunion site, which records vital oral history, with important things to tell us, and you must spend some time there. Today, if you can.]

Chamberlin also said (at the reunion) that Codd was slightly distant from the System/R project, in case it failed. And Codd would probably, he adds, have judged anything a failure against his own exacting 12 rules.

Codd's 12 rules expanded into 300, and were last seen heading north at over a thousand.

So who's to say who was right?

Commercial

The first commercial RDBMS was Multics Relational Data Store, from Honeywell, which shipped in 1976.

But at the University of California, Berkeley in 1973, Michael Stonebraker and Eugene Wong used published information on System/R to begin work on their own relational database. Codd's work - and the early published information on System/R - helped push their efforts along. Their Ingres project would eventually be commercialized by Oracle Corp., Ingres Corp. and other Silicon Valley vendors.

Codd was born in Portland, Dorset, attended Oxford University and served in the RAF in the Second World War. He joined IBM in 1953, and retired on 1 May, 1984. He subsequently formed a company with RDBMS expert Chris Date.

Jim Gray told us by email that a wake will probably be held in the San Francisco. It hasn't been organized yet, but "is likely to happen at IBM on Ted's Birthday in August." ®

Good Links

System/R reunion site
Multicians.org
any more? - ao