A framework for autonomous PostgreSQL

Self-driving, or autonomous, databases are an urgent topic as PostgreSQL becomes more and more popular. Organisations adopting PostgreSQL are facing major challenges:

• More and more development teams do not include trained Development DBAs, which means that the teams have limited abilities to design high-performance schemas, queries, and indexes, or to address performance issues resulting from the use of ORMs or vibe-coding tools.

• PostgreSQL adoption has outstripped the supply of qualified operational DBAs who can support PostgreSQL-based systems after go-live. This means the lack of talent is hampering innovation.

• Maintaining and operating databases at scale – 1000s of instances – is challenging and costly. Quarterly security updates alone have become a near-impossible task, without talking about continuous performance management, backup monitoring, etc.

This talk reviews the current work on automatic performance tuning and autonomous databases in the PostgreSQL ecosystem (including Kubernetes and AI) and puts it into the five-level framework proposed by SAE International for autonomous vehicles. We aim to evaluate existing solutions within the PostgreSQL ecosystem, highlighting those that demonstrate the most potential.