We Wish: Jim Gray's Accomplishments, Reconsidered

The Jim Gray Tribute on the UC Berkeley campus last weekend was instructive to me in several ways.

First of all, it made me realize how relatively small the "database science" community really is. Had you vaporized Zellerbach Hall at 10am last Saturday, you would have set data management back a hundred years. To me, it seemed everyone there knew each other from some past academic or business relationship.

Second, once cannot overemphasize the magnitude of Jim Gray's accomplishments. Before Jim Gray's work, database transactions were poorly understood, poorly engineered, inefficient, and expensive. (As Bruce Lindsay pointed out, in 1969, a single ATM transaction cost a bank $5. Today, it costs half a cent.)

Jim Gray (and others, but principally him) not only invented the transaction theories that serves as the "philosophical" underpinnings of OLTP (eg, ACID properties, sequential processing, concurrency, etc.) but also designed their implementation within relational database systems - as well as the benchmarks for measuring their performance. In that sense, Gray was equally influential as a theorist and engineer (and as I learned at the tribute, also supremely collaborative).

Consider that fact for a moment. Before Gray's work, crediting a bank account or booking a reservation was an inexact process. After Gray's work, it is now a reliable, efficient, precisely engineered one. (This accomplishment garnered him a Turing Award, the Nobel of computer science.) It would not be over-dramatic to say that Jim Gray made modern business systems - made e-business - possible.

So here we are today, debating the fate of Twitter, of the portable social graph, of business models for social networking. People, please. We are like ants scurrying under the feet of giants. Compared to the problems tackled by Gray - and others like him; he was not alone - these issues are trivial in the extreme. (But we'll continue to talk about them, because it's fun.)

Finally, the paradox of Gray's fate - his disappearance literally into thin air - is a cruel one. Gray himself, whose greatest accomplishments were about removing all ambiguity from transactions, is like a transaction lost in mid-flight. His family and friends are now left to cope with the challenge of "ambiguous loss", as one speaker put it, forever.

Update (June 3): Archived video of this event is now available here.


"Before Gray's work, crediting a bank account or booking a reservation was an inexact process" Might be interesting to ask Air France for example how "inexact" their online booking system was back in the mid 70s? Because sure as heck they were doing it and no one got lost. Same goes for example for the Bunker Ramo bank tellers all over Europe in the late 70s. Jim was a great engineer and scientist and I pointed that out a long time ago in my blog when he disappeared. The man was beyond genius! But he was not by any stretch of the imagination the "father" of transactional processing.

Posted by Noons on June 02, 2008 at 07:04 AM PDT #

My point is that the process itself was not based on consistent standards, and it was not inefficient (why else such expense?) If Gray was not the "father" of transaction processing (some would disagree), he was certainly its midwife.

Posted by Justin Kestelyn on June 02, 2008 at 08:45 PM PDT #

Post a Comment:
  • HTML Syntax: NOT allowed