How do you turn a jumble of vector embeddings into clear, human-readable answers? In this post, we walk through the journey from raw embeddings to precise, context-aware responses—bridging the gap between data and understanding.
I had the pleasure of speaking at PGConf.dev in Montreal, where I delivered a session on Columnar Data Store—a topic that’s quietly transforming the way we handle large-scale analytics in PostgreSQL. In this post, I’ll walk you through the core ideas from my talk and explain why columnar storage isn’t just a performance boost—it’s a game changer for modern data workloads.
pg_createsubscriber is a tool that was introduced in PostgreSQL 17 to simplify and speed up the creation of a logical replica by converting an existing physical standby server. In this blog post, we’ll explore the motivation behind the introduction of this tool, how to use it, and how it is implemented, including its internal workings.
At this year’s PGConf.dev, the premier gathering for PostgreSQL contributors, developers, and community leaders, Zhijie Hou and I had the opportunity talk about the challenges and solutions around conflict handling in logical replication — a topic increasingly relevant as PostgreSQL adoption continues to grow.
When working with embeddings in PostgreSQL, particularly for use cases like semantic search, recommendation systems, or retrieval-augmented generation (RAG), how you prepare and ingest data matters just as much as how you query it.
At first glance, storing and querying embeddings in PostgreSQL may seem impractical—but with the right setup, it’s both efficient and effective. This post covers how to design your schema, store vectors properly, and perform fast similarity searches without the usual headaches.
Embeddings are the foundation of vector search, allowing us to represent meaning-rich content like documents or queries as numerical vectors. But to use them effectively, it’s essential to understand what’s actually being embedded—whether that’s individual words, full sentences, or larger chunks of text.
As vector search becomes a foundational feature in modern applications—from semantic search and recommendation engines to AI-driven insights—developers are increasingly adopting PostgreSQL with the pgvector extension. However, one concept often creates confusion: the difference between similarity and distance.
At PGConf India 2025, I shared strategies for upgrading PostgreSQL replication clusters with no disruption to operations—highlighting examples and the evolving capabilities of logical replication.
For many organizations, the race to adopt AI is already underway. But too often, projects stall, not because of the models, but because the data underneath them is chaotic, inconsistent, or poorly understood.
A recent cryptomining attack on PostgreSQL servers revealed how misconfigurations and weak credentials can be exploited to deploy fileless malware, underscoring the need for proactive, layered security measures to defend against evolving threats. Nishchay Kothari explains.
Earlier this year, I had the pleasure of speaking at PGConf India 2025, where I presented on one of the most exciting advancements in PostgreSQL 17—how logical replication is now more resilient and failover-ready with failover logical slots—and today, I’m excited to share the slides from that talk with you.
In an era where fraudsters continually evolve their tactics, ensuring your fraud detection architecture is resilient and adaptable isn't just beneficial—it's essential. Discover how Fujitsu Enterprise Postgres 17 SP1 equips you with the tools to stay ahead.
Earlier this year, I had the incredible opportunity to present at PGConf India, where I delved into the intricacies of Write-Ahead Logging (WAL) in PostgreSQL. My presentation aimed to demystify this crucial database feature that ensures data integrity and enhances performance.
Fujitsu has participated in the PostgreSQL community for over 20 years and has contributed to PostgreSQL development by developing new features, as well as creating and reviewing correction patches. And last year we took our team contributions to a whole new level.
