David Lomet - Dedication to Jim Gray.pdf

盖国强

144

3页

0次

2024-01-21

5墨值下载

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 willing-

ness 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 contri-

butions to database and transaction processing research and technical leader-

ship 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 concur-

rency 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 ex-

hibit 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 hon-

oring 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 relational

Journal of Universal Computer Science, vol. 13, no. 8 (2007), 1044-1046

submitted: 7/7/07, accepted: 7/7/07, appeared: 28/8/07 © J.UCS

model 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) Mi-

crosoft’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 won-

derful and useful demonstration called TeraServer, demo’d last year at SIGMOD

1045

Lomet D.: Dedication to Jim Gray