Enhancing PostgreSQL Sharding Through Agentic AI: Scalability, Optimization, and Autonomous Management
Presented by:
Premnath Jangam
I am a database administrator with 8 years of experience in Postgres. Has a fashion of Researching and presenting the talks on wider community of postgres.
No video of the event yet, sorry!
PostgreSQL, a robust open-source relational database management system, has increasingly adopted Sharding techniques to address the challenges of horizontal scaling in high-volume, distributed environments. Sharding involves partitioning data across multiple nodes based on criteria such as hash keys or ranges, enabling improved performance, fault tolerance, and resource utilization. Extensions like Citus and built-in features in PostgreSQL 15+ facilitate declarative partitioning and foreign data wrappers for federated queries, yet traditional Sharding faces limitations in dynamic workload adaptation, rebalancing overhead, and schema evolution. This abstract explores the transformative potential of agentic AI—autonomous systems capable of perceiving environments, reasoning over goals, and executing actions—in expanding the scope of PostgreSQL Sharding. Agentic AI can enable intelligent, real-time shard allocation by analyzing query patterns, data skew, and node health through reinforcement learning models, minimizing downtime during re-sharding operations. Furthermore, multi-agent frameworks could orchestrate collaborative tasks, such as predictive maintenance for shard replicas or anomaly detection in distributed transactions, integrating with tools like PG bouncer for load balancing. The scope extends to hybrid AI-database architectures, where agents leverage large language models for natural-language-driven schema optimizations or automated migration from monolithic to sharded setups.
- Date:
- Duration:
- 50 min
- Room:
- Conference:
- Postgres Conference: 2026
- Language:
- Track:
- Ops
- Difficulty:
- Medium