Dedication to Jim Gray
David Lomet
(Microsoft Research, USA
lomet@microsoft.com)
As most readers undoubtedly know, Jim Gray did not return from a
sailing trip he took in January of this year (2007) and has been
missing since then, despite one of the most comprehensive and far
reaching search operations that the world has ever seen. While the
search effort was unprecedented, the willingness of people to
volunteer their time to help in the effort was not surprising given
both Jim's manifest and widely known technical accomplishments and the
many people he has befriended, helped, and mentored over the
years. Thus, we dedicate this volume, the Dagstuhl Seminar on
"Atomicity: A Unifying Concept in Computer Science" to Jim Gray, whose
work has done so much to shape our notion of atomicity and the many
roles it can play, does play, and should not be expected to play in
computer science, all topics on which Jim has published.
In 1998 Jim was awarded the ACM Turing Award for "For seminal
contributions to database and transaction processing research and
technical leadership in system implementation." More information about
Jim and the award is available on the ACM web site http://awards.acm.org/citation.cfm?id=
7232067&srt=all&aw=140&ao=AMTURING.
The complete Turing Award citation reads: "For fundamental
contributions to database and transaction processing research and
technical leadership in system implementation from research prototypes
to commercial products. The transaction is the fundamental abstraction
underlying database system concurrency and failure recovery. Gray's
work led to the definition of the desired key transaction properties:
atomicity, consistency, isolation and durability; and his locking and
recovery work demonstrated how to build database systems that exhibit
these properties. His fault tolerance work explained how to use
transactions to achieve high system availability."
At the 1999 SIGMOD Conference, I gave a short speech at a ceremony
honoring Jim and his winning the Turing Award. I'd like to reproduce
that here as it hints at both the breadth of Jim's technical
accomplishments and his influence in our technical community.
`It is a real pleasure for me to have an opportunity to speak here. It
is rare that one of our database colleagues is recognized with a
Turing Award. Indeed, it has been twenty years since Ted Codd won the
award. I think it is particularly fitting that it is Jim who has won
the award. While Ted contributed the relationalmodel to our field, Jim
provided us with the other great abstraction of our field,
transactions. These two abstractions have been, and continue to be,
central to our field.
My personal interactions with Jim go back almost 25 years, to when we
were both at IBM (though we worked on opposite sides of the
country). I can indeed remember when Jim was not a technical leader of
our field, it was back before there was a field called relational
databases, back before System R. System R was the progenitor of all
subsequent relational database systems. It showed the world that
relational database systems were REAL. As in so much of the subsequent
history of our field, Jim was a technical leader of that effort,
particularly in what became known as the RSS (relational storage
subsystem) component. I recently had the opportunity, via invitation
from Rick Snodgrass, to contribute my thoughts on a most influential
paper. For me the choice was real easy. It was Jim's "Notes on
Database Operating Systems". This paper grew out of Jim's experience
with System R and served as an early bible on how to implement
transaction support in database systems.
Later, when Jim was thinking of leaving Tandem, I played an
enthusiastic if minor role in helping to recruit him to DEC. I can
remember how we did this. First, we had to convince Jim that our
company was serious about databases. That was sometimes hard. Then we
had to convince our company that Jim should have a lab in San
Francisco. That usually took some time. But it worked, and Jim made an
enormous contribution to DEC's prominence in the TP/DB world. The most
public part of Jim's impact while at DEC was AlphaSort, demonstrating
that the Alpha chip's blazing clock time could be converted into
blazing data processing performance. But I think Jim's contribution
inside of DEC was more important. At a point when DEC's production
systems strategy was bordering on catastrophe, Jim picked up the
pieces, and turned it into a coherent technical AND business
strategy. And when DEC was withdrawing from the software arena, Jim
was a leader in the effort to keep that business going.
After DEC exited from the production software business, fate
arranged for me to be hired at Microsoft. So, let me now repeat part
of the previous story-I played an enthusiastic if minor role in
helping to recruit him to Microsoft. I can remember how we did
this. First, we had to convince Jim that our company was serious about
databases. That was sometimes hard. Then we had to convince our
company that Jim should have a lab in San Francisco. That usually took
some time. But it worked.
And once again, Jim made an enormous contribution to (this time)
Microsoft's prominence in the TP/DB world. Jim was instrumental in
bringing data cube support to SQL Server, he showed SQL Server
scalability via a wonderful and useful demonstration called
TeraServer, demo'd last year at SIGMOD and subsequently live on the
web ever since.
Jim consults with the NT
group on clusters, and the SQL Server group on concurrency control and
recovery. In my meeting with Bill Gates on his SIGMOD 98 speech, I
remember Bill saying, when asked about some database technology, that
Jim had told him about the technology, and if Jim had said it, Bill
had great confidence that this was correct. (It is not easy to impress
Bill Gates-but Jim has.)
Finally, Jim's research does not consist only of clever solutions to
hard problems. What really distinguishes Jim's work is the number of
times in which he has defined what the hard problems were: recovery,
transaction commit, concurrency control, why systems fail, sorting,
benchmarking (he was Anon of Anon et al), OLAP (data cube). Jim always
seems to be working in the future. It was Jim who defined "petabytes"
for me, when I was still trying to remember what "terabytes" meant.
I believe that giving Jim the Turing Award not only honors Jim, Jim
brings honor to the Award.'
So this dedication is truly fitting- and honors us while it honors him.
|