Michael Stonebraker

Michael Stonebraker: The Serial Founder of the Database Industry

Most database systems running on the planet today, from the free and open-source PostgreSQL to a chain of commercial products built by rival vendors, trace some piece of their lineage back to code Michael Stonebraker wrote at Berkeley in the 1970s. He is unusual among computing's most decorated researchers for having spent his entire career doing the same thing over and over — building a working database system, releasing it, then founding a company to sell it — and having it work nine separate times.

From New Hampshire to Berkeley

Michael Ralph Stonebraker was born October 11, 1943, and grew up in the small town of Milton Mills, New Hampshire. He earned a bachelor's degree in electrical engineering from Princeton University in 1965, then a master's in 1967 and a Ph.D. in 1971, both from the University of Michigan. He joined the faculty at the University of California, Berkeley, where his career-defining work began in earnest.

Proving the Relational Model Could Actually Run

In 1970, IBM researcher Edgar F. Codd published a paper proposing the relational model of data — organizing information into tables linked by shared values rather than the rigid hierarchical structures then in use — but Codd's paper was theoretical, with no working system to prove the idea was practical at scale. In 1973, Stonebraker and fellow Berkeley professor Eugene Wong set out to build one. The result, INGRES (Interactive Graphics and Retrieval System), became one of the first working implementations to demonstrate that relational databases were not just mathematically elegant but genuinely efficient to build and run, incorporating innovations in B-tree indexing, data replication, and trigger-based integrity checking that remain standard techniques in database engineering today. Stonebraker co-founded Relational Technology, Inc., later renamed Ingres Corporation and known today as Actian, with Wong and colleague Larry Rowe to commercialize the system — the first of what would become a long pattern in his career.

POSTGRES and Its Descendants

Rather than rest on INGRES's success, Stonebraker turned to what he saw as the relational model's limitations, launching the POSTGRES project to extend databases with support for complex data types, rules, and object-oriented features the original relational model had not anticipated. POSTGRES's open-source descendant, PostgreSQL, became one of the most widely deployed database systems in the world, still under active development decades after Stonebraker's original design. He commercialized a version of the technology as Illustra, later acquired by Informix, where he served as chief technology officer.

A Second Act at MIT

In 2001 Stonebraker moved to MIT, where, rather than slowing down, he opened an entirely new chapter of database research aimed at the specialized workloads that a single general-purpose relational engine could not efficiently serve. He led the development of C-Store, a column-oriented database optimized for analytical queries, commercialized through the company Vertica; H-Store, an in-memory transactional database commercialized as VoltDB; SciDB, a database built for large-scale scientific array data, commercialized through Paradigm4; and later Tamr, a data-curation company, and Hopara, among others. Across his career he founded or co-founded roughly nine companies built on his research, a rate of translating academic database theory into shipped commercial products that has few parallels in computer science.

The Turing Award

In 2014, the ACM awarded Stonebraker the A.M. Turing Award, often described as computing's Nobel Prize, "for fundamental contributions to the concepts and practices underlying modern database systems." The award, which carries a $1 million prize funded by Google, recognized not a single invention but the pattern itself: Stonebraker's demonstrated ability to identify a database architecture the field needed, build a working system that proved the concept, and release the code — often as open source — so the underlying ideas could spread through the industry regardless of which particular company he had founded to sell them.

Recognition

Beyond the Turing Award, Stonebraker received the inaugural SIGMOD Edgar F. Codd Innovations Award, named for the very theorist whose ideas he had made practical, the IEEE John von Neumann Medal, and the 2015 Commonwealth Award. He was elected an ACM Fellow in 1994 and to the National Academy of Engineering in 1997. He holds the title of professor emeritus at UC Berkeley and adjunct professor emeritus at MIT's Computer Science and Artificial Intelligence Laboratory.

Why Michael Is Called a Genius

Stonebraker's case for the label is unusually well-documented because the Turing Award citation is explicit about what it is honoring: not one clever algorithm but a sustained, repeated ability to see which database architecture the industry would need next, years before the need became obvious, and then actually build and ship it rather than merely publish about it. Colleagues in the database research community routinely describe him as the field's most consistently prescient system builder, citing the sheer number of distinct architectures — relational, object-relational, columnar, in-memory, array-based — he identified early and proved out with working code. The honest complication is narrower than with some collaborative laureates: Stonebraker's Turing Award was his alone, unshared, though virtually every project he built involved substantial co-founders and co-authors — Eugene Wong and Larry Rowe on INGRES, teams of MIT students and colleagues on C-Store and H-Store — and none of his systems succeeded as solo efforts. His genius, if the word applies, was as much architectural foresight and entrepreneurial persistence as pure theoretical originality.

Legacy

The database industry today is, in large part, an argument Stonebraker has been having with himself for fifty years about which architecture fits which workload, with each new answer spun out as its own company. PostgreSQL alone, free and still growing in adoption decades after its Berkeley origins, would be legacy enough for most careers; Stonebraker built eight more systems around it. What distinguishes his body of work from a typical academic publication record is that so much of it survived contact with industry: PostgreSQL underlies products from multiple major technology vendors, and the columnar and in-memory architectures he pioneered at MIT are now standard categories that any enterprise database buyer recognizes, regardless of whether they can name the professor who first built a working version of each one.

Achievements

• Turing Award — 2014
• Notable work: Ingres
• Affiliated with University of California, Berkeley
• Educated at Princeton University, University of Michigan and The Governor's Academy
• Worked as computer scientist