Don't Believe the Big Database Hype, Stonebraker Warns

Mike Stonebraker, a prominent figure in database design, warns against the hype surrounding new database technologies in his recent paper co-authored with Andrew Pavlo. Stonebraker, known for his critical assessments, argues that while some innovations in the last two decades have merit, many are not beneficial and will remain niche. He declares that MapReduce, particularly Hadoop, is effectively obsolete, and that key-value and document stores are either evolving into relational systems or facing challenges from them. The authors emphasize the enduring strength of the relational model and SQL, which they believe will continue to dominate despite the emergence of various database types, including columnar and vector databases. They highlight the cloud's transformative impact on database architecture, advocating for cloud-native solutions and the rise of lakehouses as successors to traditional data warehouses. Stonebraker and Pavlo also critique blockchain databases as a fading trend, noting their inefficiency for legitimate business applications. They conclude by encouraging the database community to build on historical successes while fostering innovation, suggesting that the relational model will remain central to future developments in database technology.

- Mike Stonebraker critiques the hype around new database technologies, asserting many are not beneficial.

- He emphasizes the enduring relevance of the relational model and SQL in database management.

- The authors highlight the cloud's significant impact on database architecture and the rise of lakehouses.

- They declare MapReduce, particularly Hadoop, as effectively obsolete.

- Blockchain databases are viewed as a waning trend with limited practical applications.