Michael Stonebraker

What they built

Co-designed Ingres (UC Berkeley, 1970s) and Postgres (Berkeley, 1980s) — the academic lineages behind nearly every modern relational database. Later founded or co-founded Illustra, Cohera, StreamBase, Vertica, VoltDB, SciDB, Tamr, and DBOS. 2014 ACM A.M. Turing Award laureate. Famous for periodically writing papers arguing his own prior architecture is obsolete.

In their own words

• "The last 25 years of commercial DBMS development can be summed up in a single phrase: 'One size fits all'." — "One Size Fits All": An Idea Whose Time Has Come and Gone, ICDE 2005
• "Ingres was always technically better and Postgres was practically better. It's more flexible, and it's open source... closed source databases are not the wave of the future and I think Oracle is highly priced and not very flexible." — The Register, Dec 2023
• "When MySQL was bought by Oracle, developers got suspicious in droves, and defected to PostgreSQL. It was another happy accident." — The Register, 2023
• On DBOS: "Basically, the operating system is an application to the database, rather than the other way around." — The Register, 2023
• "I would have a hard time picking one of these four systems as the most satisfying; it's like asking which of my children I love the most." — on his major systems
• "I can't imagine playing golf three days a week. I like what I do, and I will do it as long as I can be intellectually competitive." — The Register, 2023 (he was ~80)

Principles as they articulated them

• No single DBMS architecture can win. The market will (and should) fracture into specialized engines — column stores for analytics, main-memory for OLTP, array stores for science, stream processors for events — possibly sharing a front-end parser but not a storage engine.
• Row stores are the wrong default for analytics. His C-Store / Vertica work argued column orientation is 10–100× faster for read-mostly workloads; the industry eventually agreed.
• Main memory changes everything. H-Store / VoltDB: if your working set fits in RAM, the legacy OLTP architecture (buffer pools, locking, WAL) is mostly overhead. Strip it.
• Open source wins the long game. He has consistently argued closed-source databases are structurally doomed and repeatedly released his research systems under permissive licenses.
• Kill your own darlings. He has publicly argued each of his own systems is obsolete once the next one is ready — a posture more typical of a researcher than a founder.

What surprised me in research

• Stonebraker is arguably the only founder who has systematically argued against his own previous products in print. The "One Size Fits All" papers (2005, 2007) directly attack the universal-DBMS premise that his own Postgres helped establish. This self-revision is the pattern of his career, not an exception.
• Postgres's survival is, by his telling, a happy accident: he treats the MySQL/Oracle acquisition as the real cause of Postgres dominance — a developer flight, not an architectural victory. He is unsentimental about it.
• His current project DBOS inverts a 50-year assumption: instead of the database being an application running on the OS, the OS runs on top of the database. Application state, scheduler state, file metadata — all become tables in a distributed DBMS. It is a genuine architectural provocation from someone who could easily be retired.
• In the 2023 Register interview he was still working full-time at ~80, explicitly because he cannot stomach the idea of not being "intellectually competitive." This is the dominant note in his recent public appearances.
• The famous metaphor is children, not killing — "it's like asking which of my children I love the most" — but the pattern looks like serial obsolescence from the outside.

Recent or later work

• DBOS (2022–present): co-founded with Matei Zaharia and others. Commercialized as DBOS Inc. The underlying research paper ("DBOS: A DBMS-oriented Operating System," VLDB 2022) argues for a full OS-as-database rearchitecture.
• Tamr (2013–present): data integration/unification at scale; one of his few non-engine companies, attacking the "dirty data" problem.
• Continues to co-teach and publish at MIT CSAIL. Received the ACM SIGMOD Systems Award and the 2014 Turing Award for "fundamental contributions to the concepts and practices underlying modern database systems."
• ACM published Making Databases Work: The Pragmatic Wisdom of Michael Stonebraker (2018) — a Festschrift collecting his key papers and commentary.

Sources

• https://amturing.acm.org/award_winners/stonebraker_1172121.cfm — Michael Stonebraker, A.M. Turing Award Laureate — amturing.acm.org
• https://cs.brown.edu/~ugur/fits_all.pdf — "One Size Fits All": An Idea Whose Time Has Come and Gone — cs.brown.edu
• https://www.theregister.com/2023/12/26/michael_stonebraker_feature — Postgres pioneer promises to upend the database once more — theregister.com
• https://thenewstack.io/dr-michael-stonebraker-a-short-history-of-database-systems/ — A Short History of Database Systems — thenewstack.io
• https://dl.acm.org/doi/book/10.1145/3226595 — Making Databases Work (ACM Books, 2018) — dl.acm.org
• https://www.dbos.dev/ — DBOS — dbos.dev