Our year-end Hyderabad PostgreSQL User Group Meetup #12 was successfully completed with an overwhelming turnout of 120 attendees and 8 amazing speakers. It was truly an awesome treat to close the year on such a high note.
A heartfelt thank you to Entain for the support that made this meetup possible.
Let’s catch up next year with the same energy and enthusiasm at Meetup #13!
Neeta Goel | Newt Global 
Ramanathan Natarajan| Newt Global
Achieving performance parity or improvement while migrating from from Oracle to PostgreSQL goes beyond mere functional translation. Speakers will share real life examples of different solutions in the context of workload characteristics for Oracle specific features such as Global Package Variables, PRAGMA AUTONOMOUS_TRANSACTION which are not supported in PostgreSQL. Understanding the differences in EXPLAIN PLAN and how to use it to find the performance root causes as well as common issues due to difference in how table partitions, index column order and many other features work in PostgreSQL.
Furthermore, we’ll discuss implications of application layer and the architecture, for example how Object-Relational Mappers (ORMs) can inadvertently create performance issues, discussing common anti-patterns like N+1 queries and strategies to mitigate them.
Finally, we’ll cover the essential role of tuning PostgreSQL server parameters & vacuum settings based on workload
Amit Jain | JP Morgan Chase & Co
Shameer Bhupathy | OpenSource DB
Gone are the days when scalability was an afterthought — it now drives every architectural decision.
PostgreSQL’s built-in partitioning has come a long way, now providing a mature and efficient framework for organizing and optimizing massive, rapidly expanding datasets.
In this session, we’ll explore the full spectrum of PostgreSQL partitioning — starting with core principles and partitioning methods, then moving into advanced techniques and hard-earned insights from real-world production environments.
We’ll trace the evolution of partitioning in PostgreSQL, demystify key enhancements like better partition pruning and join support default partitions, skip scan on multicolumn Indexes and global indexes, and explore practical patterns that drive performance and maintainability in production environments.
Key Takeaways:
– Grasp the why and how of PostgreSQL partitioning
– Apply scalable, high-performance partitioning strategies
– Get real-world examples and lessons learned from deploying partitioning in live systems
– Gain practical insights to design or optimize systems with partitioning in mind
Akshat Jaimini | DeepSource
This talk will take a deep dive into how PostgreSQL implements Write Ahead Logging. We’ll begin by delving directly into the PostgreSQL source code to uncover how WAL files are generated, written, and managed under the hood. From low-level internals to high-level design decisions, we’ll trace the full lifecycle of a WAL record. The session will then shift to real-world use cases that highlight the power and flexibility WAL provides, exploring benefits like performance optimization, crash recovery, usage in streaming replication and point in time recovery. This talk will help attendees in understanding the importance of WAL and appreciate the silent warrior of durability.
Shreya Aithal | Microsoft
Everyone’s building AI Agents. But ask what they really are, and the answers get fuzzy.
Under the broader umbrella of “a system that perceives its surroundings, gathers information, and uses it to perform actions and complete tasks autonomously,” we’ve seen a shift from traditional rule-based agents to learning-based agents, and now to multi-agent architectures. But what do these types of agents actually mean? How do you decide which one fits your use case? How do you build multi-agent systems that solve real-world problems? And are they truly necessary, or can Machine Learning and Generative AI apps do the job?
Join this session to gain a solid foundational understanding of AI Agents and their relevance in today’s world. You’ll learn what each type of agent entails, supported by live demos and practical examples. We’ll explore the current landscape of AI Agents, dive into frameworks like LangGraph, LlamaIndex, and Semantic Kernel (with a detailed feature-by-feature comparison), and show how to integrate them with PostgreSQL. You’ll also get clarity on today’s evolving terminologies and standards – all through a comprehensive end-to-end demo!
Consider this your complete beginner’s guide to AI Agents.
Aswini Kumar Tummak | AWS
Sai Krishna Namburu | AWS
PostgreSQL 18 introduces key advancements in performance, observability, and security—making it more capable than ever for enterprise workloads. This session explores the new asynchronous I/O subsystem, optimizer enhancements, virtual generated columns, and OAuth integration. Attendees will gain practical insights into adopting these features and optimizing PostgreSQL 18 for modern applications.
Key Takeaways:
1. Understand the Impact of Asynchronous I/O
Learn how PostgreSQL 18’s new async I/O subsystem improves performance for I/O-bound workloads and what configuration changes are needed to leverage it effectively.
2. Explore Query Optimization Enhancements
Gain insight into how the optimizer now handles self-joins, partitions, and common expressions more efficiently resulting in faster and more predictable query plans.
3. Leverage Virtual Generated Columns and Advanced Replication
Discover the benefits of virtual generated columns as default, and how enhanced logical replication (including support for generated columns and parallel streaming) simplifies distributed deployments.
4. Enhance Observability with New Monitoring Tools
Learn how to use `pg_stat_io`, updated `EXPLAIN ANALYZE`, and other observability improvements to diagnose and tune system performance with greater accuracy.
5. Modernize Security with OAuth Integration
Understand how native OAuth support enab
Ragesh Hajela – Fujitsu
Chiranmoy Bhattacharya – Fujitsu
As Arm CPUs become increasingly prevalent in data centers and high-performance computing, optimizing core database operations for vector performance is critical. To enable PostgreSQL to take full advantage of modern Arm architectures, we contributed the first-ever integration of Scalable Vector Extensions (SVE) into PostgreSQL. This work delivers a portable, vector-length-agnostic SIMD backend that boosts performance across a wide range of SVE-enabled Arm CPUs.
Our initial patch introduced vectorized versions of critical routines and setting the foundation for broader SIMD acceleration in PostgreSQL. First contribution introduced SVE-accelerated pg_popcount, used in visibility map operations that are fundamental to ANALYZE and VACUUM commands. This change delivered up to 7x speedup for that function on SVE-enabled platforms. Our second contribution focuses on optimizing hex encoding/decoding for bytea data types using SVE, achieving up to 10x speedup for these routines. Together, these improvements contributed upto ~11% performance gain on few queries in the TPC benchmarks.
These kernels are written using Arm SVE intrinsics and designed to scale with vector lengths from 128 to 512 bits, enabling performance portability across CPUs like AWS Graviton, A64FX, NVIDIA Grace, and Fujitsu’s upcoming MONAKA Processor (SVE2, 256-bit).
This talk will detail the motivation, design, and implementation of SVE support in PostgreSQL, discuss upstream collaboration, and present early performance metrics. We’ll also explore how this vectorization effort aligns with PostgreSQL’s long-term goals for extensibility and multi-architecture performance optimization.
InduTeja Aligeti | AWS
Bhavani Akundi | AWS
As organizations seek to embed intelligence directly into their applications, PostgreSQL has emerged as a powerful and flexible data platform for AI/ML innovation. This session explores how PostgreSQL extensions like pgvector, PostgresML, MADlib, and plpythonu can be harnessed to build AI-powered applications natively within the database. Attendees will learn how to store and search vector embeddings, run machine learning models inside Postgres, and integrate seamlessly with modern AI toolchains. Through real-world use cases and live demos, we’ll demonstrate how to minimize data movement, improve performance, and simplify architecture by keeping AI close to where data lives—inside PostgreSQL.
The talk will cover:
Practical use cases: semantic search, recommendation systems, anomaly detection.
Architecture patterns for reducing data movement and latency.
Benefits of a unified AI + database approach in terms of cost, performance, and simplicity.
Ameen Abbas | Velocis Systems Pvt Ltd
Rajesh Madiwale | AWS
We’ll start with an overview of PostgreSQL’s key configuration parameters, understanding their role in memory management, query processing, and background operations. You’ll learn how to assess your current workload and adjust memory-related settings like shared_buffers, work_mem, and effective_cache_size to align with your system resources and performance goals.
Whether you’re running PostgreSQL on bare-metal, cloud VMs, or containers, this workshop will equip you with practical strategies, pro tips, and hands-on examples to transform your instance from a generic setup to a finely tuned machine.
What You’ll Learn:
– When and why to override default settings
– Memory tuning best practices
– Autovacuum configuration strategies
– Performance tuning methodology
– Real-world case studies and tuning checklists
Padma MP| Aiven
Pranay Shree| Aiven
This talk shares real-world insights from managing thousands of PostgreSQL instances in the cloud. We’ll cover the insights on scaling reads, writes, and connections, how to design for resilience, and what monitoring really matters. You’ll walk away with practical tips, lessons at what it takes to run PostgreSQL reliably at scale — whether you’re on Aiven or rolling your own.
Mukesh Marodia | Google 
Virendra Singla | Google
Unearth the secrets to optimal PostgreSQL performance in this interactive, hands-on workshop! Guided by expert participants will become performance detectives, working in teams to solve real-world database challenges presented as a thrilling treasure hunt. Each clue will lead you deeper into key optimization areas, including:
Mastering Query Plan Analysis
Investigating Resource Wait Events
Identifying System Host Metrics & Bottlenecks
Implementing Advanced Index Strategies
Optimizing Buffer & Memory Allocation
Understanding Table Statistics & Autovacuum Best Practices
Performing Effective Workload Analysis & Benchmarking
This workshop is an invaluable experience for DBAs and developers seeking to significantly enhance their PostgreSQL troubleshooting and tuning skills. Join the hunt and unlock your database’s full potential!
Bikash Rout | AWS
Sashikanth Pattanaik | AWS 
PostgreSQL is a powerful open-source database, but it is true potential is often underused due to lack of awareness around its built-in capabilities and ecosystem tools. This session focuses on the minimum essential features, tools, extensions, and best practices that can significantly improve the day-to-day productivity of both developers and DBAs. Whether you’re building applications, writing queries, or managing database infrastructure, this session gives you the practical knowledge needed to work smarter and faster with PostgreSQL.
Agenda:
1. Core PostgreSQL Features Every Dev/DBA Should Master
– SQL productivity tips: RETURNING, UPSERT, CTE, window functions
– Writing efficient PL/pgSQL functions and using RAISE, EXCEPTION
– Role and permission basics for secure access
– Understanding EXPLAIN & ANALYZE for query insights
2. Must-Know Extensions (Use Cases + Demos)
– Query performance tracking & tuning
– Case-insensitive search support
– Semi-structured data handling
3. Tools that save time
4. PostgreSQL Security
– Roles, users, privileges
– RLS
– Security definer/invoker
– Security extensions
Jobin Augustine | Percona
There is a proliferation of extensions for PostgreSQL. There are extensions for pretty much everything possible.
Extension developers and enthusiasts often highlight and discuss extensions’ positive aspects but remain silent about the problems that they can create. As a consequence, many users remain completely unaware of the potential problems.
This presentation will have
Live demonstrations. Because seeing is convincing.
References to benchmarks and major incidents.
This talk discusses
Extensions can crash the database. – One of the biggest reasons for many of the segfaults
Extensions can cause corruption
Extensions can leak memory
Slow down PostgreSQL to a great extent
Extensions can break the Security model
Extensions can break PostgreSQL features
Conflicts between extensions
Order of extensions loaded.
Jobin Augustiine | Percona
There are many misunderstandings surrounding Availability and disaster Recovery. This talk is based on several years of real-world experience in setting up, managing, and supporting incidents and RCA investigations of HA solutions for many organisations.
The talk covers
* Read Replicas: Myths and Facts
Discusses Problems of redirecting SELECTs to the Standby side and Possible uses and solutions
* Synchronous Standby: Myths and Facts
* Discusses the myths/expectations from a synchronous standby and the facts.
* Troubles one should expect from a synchronous standby
* Application availability vs Database Availability.
Discusses how database availability is different from application availability and what matters. Key concepts like Availability 9s, MTBF, RPO & RTO
* Understanding Timeline changes and data loss
* Understanding Rewind and Data Loss
* HA vs DR – Key concepts.
* Backups and HA – What to expect
* Connection routing, Proxies and External Poolers in HA cluster
* Solutions like IP failover and their implementations
* Split-Brain – The recipe for Disaster.
Some of the “HA Solutions” don’t have any protection against Split-Brain. Understand the risk.
* Recommendations on open-source tools and frameworks are available.
Kattamuri Kowshiq | Accenture 
Why maintain five different services when one rock-solid database can do it all? PostgreSQL has evolved far beyond a traditional relational store—today, it can serve as your queue, cache, search engine, key-value store, and even power real-time features. In this talk, we’ll explore how leveraging PostgreSQL as a unified backend simplifies infrastructure, reduces costs, and speeds up development—without compromising on performance or scalability.
Through practical examples and production-tested patterns, we’ll uncover how PostgreSQL can replace Redis, RabbitMQ, Elasticsearch, and more—all while preserving ACID guarantees. If you’re building modern applications and want to move fast without overengineering from day one, this session will show why PostgreSQL might just be the only tool you need.
Aryan Sharma | Blueflag Security
A default PostgreSQL installation can be dangerously permissive. While DBAs are diligent about setting roles and policies and privileges, true database security extends far beyond GRANT. This talk dives into critical, often-overlooked configurations and attack vectors to harden your database. This is a practical, two-part guide for engineers and DBAs ready to move beyond basics.
The first part: “Bare Minimums”, is a hands-on exploration of hardening database server itself. We start with initdb, enforce encrypted connections using hostssl, prevent password leakage in logs and configure meaningful audit trails.
The second part, “When ‘Best’ isn’t enough” explores reality that a “perfectly configured” database is not always perfectly secure. It explores how application-level vulnerabilities like SQL injection or network-level attacks like ARP spoofing can bypass your security controls, potentially compromising your data.
You will leave this talk with a checklist of hardening techniques and a broader security mindset to defend PostgreSQL instances against threats.
Yogesh Jain | Enterprise DB
PostgreSQL is widely used across cloud, Kubernetes, and bare metal environments. But how can we monitor it efficiently across all these setups without vendor lock-in or high costs?
In this talk, we’ll explore how OpenTelemetry helps achieve consistent and scalable observability for Postgres – no matter where it runs. We’ll walk through how to collect metrics and logs, integrate them with open-source observability tools, and optimize monitoring for both performance and cost.
Expect live demos and practical insights on making PostgreSQL monitoring simple, efficient, and future-ready with OpenTelemetry.
Key Takeaways:
– How OpenTelemetry brings consistent Postgres observability across cloud, kubernetes, and bare metal
– How to collect telemetry data (metrics and logs) for effective Postgres monitoring
– How to use open-source tools (Prometheus, Loki, Grafana) to observe Postgres
– Tips to reduce observability costs while keeping full visibility
– Live demo of Postgres monitoring in cloud-native environments
Sivasankar Prasad – Ericsson
AI projects today often rely on high-dimensional vector embeddings—for example, from text, images, audio, or user behavior—that power semantic search, recommendation systems, and natural language processing pipelines. Traditionally, handling such vectors required dedicated vector databases, introducing complexity in data pipelines and system architecture.
This talk focuses on the benefits of using pgvector, a powerful PostgreSQL extension that enables native vector similarity search directly within the PostgreSQL cluster. By embedding AI capabilities into the same environment as your relational data, pgvector eliminates the need for ETL pipelines between systems, improves performance through locality, and maintains transactional integrity. We’ll cover real-world use cases, performance considerations, and how pgvector simplifies deploying AI features like semantic search, nearest neighbor queries, and hybrid filtering using structured + unstructured data. Ideal for data engineers, ML practitioners, and Postgres users looking to integrate AI functionality with minimal infrastructure overhead.
Kshetrapaldesi Tutika | Microsoft
In this talk, I will share how PostgreSQL has improved its security features over the years—from version 10 all the way to version 17. I’ll explain changes in user management, roles, and privileges, especially the update in version 15 where users no longer get the default CREATE privilege in the public schema. I’ll also talk about column-level security introduced in version 15, with real-life examples like protecting healthcare data.
I’ll then highlight the better role management in version 17, which gives more control and flexibility through role inheritance.
The session will also cover authentication and encryption improvements:
SCRAM-SHA-256 authentication from version 10
LDAP updates in version 11
GSSAPI encryption from version 12
Stronger password policies from version 13
SSL improvements in version 14
Direct TLS handshake support in version 17
To wrap up, I’ll talk about the new incremental backup feature in version 17, which helps prevent attackers from rebuilding the database without full backup access.
Nikhil Hegde
You run a few queries in development everything looks fast. But once your app goes live, you notice strange things: slow queries, deadlocks, growing table sizes, and confusing inconsistencies. Even something as simple as SELECT COUNT(*) feels unexpectedly slow. What’s going on?
In most cases, the answer lies in how PostgreSQL handles concurrency through MVCC—Multi-Version Concurrency Control. It’s a powerful system that lets PostgreSQL support high-concurrency workloads without locking readers and writers but it also introduces complexity that can catch you off guard.
This talk will walk through how MVCC actually works under the hood: how row versions are managed, how visibility is decided, and how background processes like autovacuum come into play. We’ll connect these internals to real-world issues you’ve probably faced—and explain how to solve them.
By the end of the session, you’ll know how to:
* Diagnose and resolve common performance issues caused by MVCC behavior
* Make informed choices around isolation levels and long-running transactions
* Tune autovacuum and avoid table bloat in write-heavy workloads
* Write safer, more scalable queries in concurrent environments
Whether you’re a developer, DBA, or architect, this talk will help you stop guessing and start making informed decisions about how your PostgreSQL database behaves under load.
Vikas Gupta | Amazon
Rajesh Madiwale | Amazon
This technical presentation explores real-world challenges encountered while operating PostgreSQL in high-stakes production environments. Through an in-depth analysis of performance bottlenecks, we examine how seemingly routine operations can cascade into significant system degradation. The discussion covers critical aspects of PostgreSQL that impact database performance.
We present a systematic approach to troubleshooting, backed by metrics and practical examples, demonstrating how proper configuration tuning and proactive monitoring can prevent common failure modes. The session includes case studies of production incidents, analyzing root causes from both architectural and operational perspectives. Special attention is given to query performance optimization, resource utilization patterns and novice ideas to tackle them.
– Scary Plans & Statistics out of way
– Ineffective Vacuum process
– Impacts of Idle connections
– Impacts of MultiXacts
– Memory Utilizations
– Rouge queries consuming 100% CPU
– Lock Contention
Lahari Giddi | OpenSource DB
Keerthi Seetha | OpenSource DB
PostgreSQL is famously reliable, but what happens when you push it beyond its comfort zone? In this talk, we unpack real-world experiments and production journeys using multi-master replication (MMR) and distributed PostgreSQL setups. We’ll explore brittle scripts, unexpected race conditions, split-brain scares, and hard-won breakthroughs. Whether you’re scaling writes, chasing high availability, or just PostgreSQL curious, you’ll walk away with practical insights, cautionary tales, and a clearer sense of how this beloved database could be Multi-master and distributed.
Vibhor Kumar | EDB 
For DBAs accustomed to Oracle’s ecosystem, the ability to surgically rewind data with Flashback is the gold standard for granular recovery. While PostgreSQL’s Point-in-Time Recovery (PITR) is exceptionally powerful for recovering an entire cluster, it often feels like a sledgehammer when a scalpel is needed. How do we recover a single table that was accidentally dropped, or reverse a fat-fingered UPDATE that affected millions of rows, without restoring the entire database? This session dives deep into the techniques and internal mechanisms that allow us to emulate Oracle’s Flashback capabilities using native PostgreSQL features. We will move beyond theory and into practice, unraveling the transaction log to perform surgical data recovery operations. |
