Conversation about Autonomous PostgreSQL: Bruce Momjian, VP, Postgres Evangelist @ EDB

At the end of May, I had the opportunity to talk to Bruce Momjian, VP, Postgres Evangelist at EDB, about Autonomous PostgreSQL and what his expectations are about AI and PostgreSQL.

Watch the full interview above — here are a few of my notes and key takeaways.

Bruce has been with PostgreSQL for over 30 years, and Bruce and I have had the opportunity to work closely together at EDB for 13 years. We touched on a wide range of topics.

We started with the historical perspective: when did performance and performance tuning become an important subject for PostgreSQL? Bruce takes us back to the reality of 1996, when PostgreSQL was a less mature code base and the focus was on stability, SQL standards compliance, and the right organizational structure to create a community, systematically fix bugs, and create reliable distributions.

Then came basic enterprise requirements like crash recovery, incremental backup, and disaster recovery. Once those basics were in place, the new hardware capabilities and workload complexities could be addressed, which also led to a significant increase in system parameters, data caching strategies, and resource management.

Bruce thinks that approximately in 2016, PostgreSQL entered the innovation phase with new capabilities such as JSON and pgvector — innovations that explain why it is the #1 database choice for developers today.

The discussion took us to exploring what's next and what role AI would play. pgvector was a huge leap forward and built on Michael Stonebraker's vision of the object-relational database that is open to being extended with new data types and operators. Bruce positions AI in the context of the hype cycle and emphasizes the opportunity of AI to democratize access to data.

We also had a short discussion of MCP versus RAG, and Bruce shared his perspective on MCP as a tool to address some of the shortcomings of LLMs, as MCP helps share factual data and meaningful context with the LLM, thus making generative AI much more meaningful.

Autonomous PostgreSQL, or AI to help the DBA, was the core part of our discussion. Bruce emphasized his opinion that AI can analyze situations and propose actions to the DBA, but the DBA is ultimately the responsible decision maker. AI plays a key role in helping the DBA filter the information, make sense of it, and identify which actions need to be taken and why. AI today can do more than just apply simple rules, especially when statistical optimization is needed to identify the best outcome.

According to Bruce, the next exciting developments in AI and PostgreSQL are probably in the area of allowing non-programmers to access a database by helping them write SQL — an area where Bruce is currently writing a talk. Democratizing data is a big topic where Bruce will spend more time over the next few months. In the core database, Bruce doesn't foresee any major new AI-focused developments, except for new features in pgvector and interesting pgvector derivatives that take advantage of GPU-based hardware acceleration. In the area of Autonomous PostgreSQL (i.e., AI for the database), Bruce sees many opportunities for health checks, index tuning, schema changes, and config parameters, and compares this to whack-a-mole, where many techniques — especially statistical ones like AI — have to be brought to bear on the problem, and where intelligent tradeoffs are needed to find a balanced optimum.

We ended the conversation with a look forward, where Bruce reflected on the last 30 years of the PostgreSQL community and project. Many of the impediments to using PostgreSQL have faded away, and the PostgreSQL community's long-term discipline and focus on the user base have paid off. As long as the community focuses on service and avoids getting embroiled in territorial or selfish approaches, PostgreSQL will always bring in new ideas — and that's what will take PostgreSQL forward.

This was the second in a series of interviews about Autonomous PostgreSQL. The first interview was with Luigi Nardi, CEO and founder of DBtune.

If you have any feedback, questions, or suggestions for future interviews, please contact us at autonomouspostgresql@dbtune.com.