PG Down Under 2025 Program

presented by PDGU Committee

presented by Thomas Munro

This is a three part talk. The first part gives a very high level overview of the areas of development in recent major releases, including planning, execution, partitioning, replication, backups, monitoring, security, I/O, ongoing modernisation work and more.

In the second part, we'll look more closely at three selected new features that bring everyday benefits in terms of usability and mai...

presented by Ravi Kiran Venkata Bhaskarabhatta

Modern day applications are increasingly expected to deliver flexibility and scalability to meet the demands of global users and unpredictable workloads. Organizations require databases that can dynamically grow beyond the boundaries of traditional CPU, memory, and storage often scaling horizontally. These applications are being redesigned to support distributed systems capable of delivering ul...

presented by Nishchay Kothari

Schema migrations in modern applications are often fraught with challenges, including downtime, client application disruption, and complexity in handling breaking changes. PostgreSQL pgroll addresses these issues by providing an innovative solution for zero-downtime, reversible schema migrations.

pgroll is an open-source command-line tool designed for PostgreSQL, ensuring that schema updates...

presented by PDGU Committee

presented by Arul Arulappan

The talk will be presented by myself and Emma Fitzgerald. We will cover: - Criticality of Postgres for Atlassian. Most of Atlassian's products use Postgres for its data store. - How our Postgres infrastructure, which is one of the largest Postgres fleet in the world, is managed today in the cloud. - Limitations/Challenges with this system. This section will focus on general challenges with ...

presented by Alena Rybakina

Pull-up is an optimization that already exists in PostgreSQL’s planner. It rewrites eligible subqueries into the main query level so that the planner can work with a unified join tree rather than treating subqueries as independent nodes.

Currently, PostgreSQL applies pull-up to several common subquery forms, including: - IN subqueries - EXISTS subqueries - ANY expressions (where applicabl...

presented by Peter Smith

PostgreSQL's CREATE PUBLICATION command now does more than just copy tables. This presentation explores how modern PostgreSQL enables fine-grained control over exactly what data gets replicated.

We'll examine the command's evolution across PostgreSQL versions, showcasing capabilities including:

• Selective table publishing - Choose exactly which tables to replicate
• Row filtering - Repl...

presented by Rajesh Kandasamy

The landscape of job scheduling within PostgreSQL has undergone a significant transformation, moving from external, administratively intensive tools to sophisticated, in-database automation engines. This evolution, marked by the progression from pgagent to pg_cron and pg_timetable, reflects a broader trend towards operational simplicity, enhanced flexibility, and tighter integration with the da...

presented by PDGU Committee

presented by Tarun Chatterjee

The rise of Generative AI has made Retrieval-Augmented Generation (RAG) a cornerstone for building intelligent, context-aware applications. While many solutions turn to specialized vector databases, this often introduces architectural complexity, data silos, and new operational overhead. What if you could build highly scalable, enterprise-grade RAG applications using the trusted, powerful, and ...

presented by Mark Wong

This is a tale about one company's experience with a database schema design refactor to use the Entity-Attribute-Value (EAV) data model on its members. I will describe the original data model and why there was a need to move to a new data model, which the EAV was chosen for. This decision was made without realizing that the EAV data model is considered an anti-pattern for relational data model...

presented by Nikhil Bayawat

In this talk, I explore the technical evolution of the PostgreSQL query optimizer over the past decade—from version 9.4 in 2015 to the upcoming PostgreSQL 18 in 2025. I examine how continuous improvements in planning, cost estimation, indexing strategies, and parallelism have helped PostgreSQL become a highly performant and trustworthy database engine for modern workloads.

I walk through key...

presented by Arpitha Vasudeva

High availability for PostgreSQL takes on new dimensions in cloud environments. This session breaks down the key architectural decisions teams face when designing for PostgreSQL high availability in the cloud. We'll explore read replicas, failover mechanisms, and backup strategies, comparing native PostgreSQL capabilities with cloud-native features. Using practical examples, we'll examine how t...

presented by PDGU Committee

presented by Roneel Kumar and Sameer Kumar

SQL Server to PostgreSQL migrations present a complex array of technical differences that can challenge even experienced DBAs - we've seen this firsthand across hundreds of hours spent with customers managing databases from tens of GiBs to several terabytes. Successfully navigating these migrations requires deep understanding of how critical components like backups, collations, and case sensiti...

presented by Vigneshwaran C

This talk will dive into a series of recent contributions to PostgreSQL, made possible by a team working with sustained focus on core database infrastructure. Rather than offering high-level overviews, the session will explore the technical design and implementation details behind several complex features we’ve delivered. These include: Conflict Detection for Update-Deleted Tuples – enhancing...

presented by Hari Kiran

PostgreSQL has long been celebrated for its rock-solid architecture and passionate Open-source community. But building a successful business on Postgres takes more than great engineering; it requires listening, translating, and advocating. Enter the Customer Success Evangelist: part technologist, part storyteller, and part trusted advisor.

In this talk, we’ll explore how companies like OpenS...

presented by Dylan Griffith

Problem: Idle Replicas

Many open source tools exist today for creating high availability deployments of Postgres. Most of these tools involve having 1 primary database and a fleet of replicas which have some small amount of replication lag. Almost all of these solutions mean running multiple Postgres servers with the same amount of disk space, CPU and memory in order to ensure that any re...

presented by PDGU Committee

presented by PDGU Committee

presented by PDGU Committee

presented by PDGU Committee