~postgresweekly | Bookmarks (121)

PGlite 0.3: Postgres on WASM, Now Based on Postgres 17 — PGlite is a WebAssembly-based build of Postgres that enables you to run an instance of Postgres wherever WebAssembly can be run (such as in the browser as in this demo). v0.3 takes a big step forward in moving up to Postgres 17.4. (GitHub repo.)

Why All Your Workflows Should Be Postgres Rows — It may be no surprise the folks behind the DBOS durable execution/workflow library think storing workflows in Postgres is a good idea, but here’s how they think about it from a schema POV.

Jepsen: Amazon RDS for PostgreSQL Multi-AZ Clusters Violate Snapshot Isolation — Jepsen is well known for its analysis of distributed systems. After some experimentation, the author found that Amazon RDS for PostgreSQL multi-AZ clusters (specifically) violate Snapshot Isolation, the strongest consistency model supported across all endpoints.

Is Your Application Prepared for Your Next High Traffic Event? — Are you tired of your Postgres database feeling like a black box? Watch Postgres expert Lukas Fittl walk through the latest pganalyze features to tune slow queries, catch regressions, and keep Postgres running at its best in this on-demand video.

What Really Happens When You Drop a Column — When you run a ALTER TABLE .. DROP COLUMN you probably don’t think too hard about what’s going on behind the scenes, but sometimes specifics of the process can rear their ugly head. In short, ‘dropped’ columns may be gone but not forgotten.

Hacking the Postgres Statistics Tables for Faster Queries — Postgres’s internal statistics drive query planning and index choices, but the process isn’t perfect. This practical post shows how to manually add extended statistics to teach Postgres about useful correlations — and how even a single CREATE STATISTICS can make a query 97x faster.

Fully Managed Postgres with Aiven for PostgreSQL® — Set up a fully managed Postgres database in 10 minutes and focus on your code – let Aiven do the heavy lifting.

Scaling Certificate Transparency with 100B+ Rows of Data — Merklemap is a service that aggregates together certificate transparency data for issued TLS certificates – here’s how they use Postgres for task of managing data spanning 20TB of storage and 100 billion rows.

Striping Postgres Data Volumes - A Free Lunch? — Depending on your usage patterns, yes, striping together cheap cloud volumes can yield significant throughput and latency benefits.

Reordering Expressions in SQL Queries for Performance — Techniques to speed up query execution, with a focus on rearranging conditions in filter expressions, JOINs, HAVING clauses, and similar constructs.

Doltgres Beta Launch: A Version Controlled SQL Database — Doltgres is a Go-powered Postgres-flavored variant of the Dolt SQL database that offers Git-style merging and forking in a Postgres-wire compatible form.

A Postgres Language Server for Your IDE — Most modern code editors support language servers for providing language-specific coding enhancements like auto-complete, syntax highlighting, and refactoring. Supabase has come up with such a server for Postgres too, offering linting, autocompletion, error highlighting and type-checking out of the box.

🤖 Postgres Pro: An MCP Server to Let AI Agents Tune and Improve Your Database Usage — I suspect it’ll need to be renamed (so as not to confused with the company of the same name?), but this is an interesting first step at providing a bridge between modern AI models and agents and Postgres by way of an MCP server that lets AI agents...

A Roadmap to Scaling Postgres — An experienced Postgres specialist presented an ordered list of steps to think through when scaling up a Postgres deployment from first getting your data model right, through to indexing, hardware, tuning, and sharding.

2x-40x Faster Docker Builds with Blacksmith — With a one-line code change, Blacksmith can make your Docker builds incremental by mounting your Docker layer cache into your GitHub Actions runner. Blacksmith is used by 350+ companies like Ashby, Clerk, and Mintlify.

📈 Making Postgres Better with OrioleDB — OrioleDB is an alternative storage engine for Postgres that extends the table access method framework to optimize performance on modern hardware. But by how much? Brian runs some tests.

Creating Histograms with Postgres — A histogram visualizes data distribution by grouping values into bins and displaying the frequency of each bin. Postgres has a width_bucket to help you do this, and Elizabeth shows off some fantastic queries here.

pg_auto_failover 2.2: Automated Failover Extension — An extension and service that monitors and manages automated failover for a Postgres cluster. Now supporting Postgres 17.

🕒 Time to Better Know the Time in Postgres — If you’ve ever been guilty of just crossing your fingers and hoping for the best when working with times in Postgres, this is for you. Packed with examples and advice, Radim looks at how different time data types work, how to work with time zones, intervals, time ranges, and more.

2025 Postgres Extensions Mini Summit Two — Several Postgres users and developers held a virtual event recently discussing recent progress in adding an ‘extension search path’ feature to Postgres, as a way to make it easier for Postgres installs to discover and pick up newly installed extensions (there’s a 50-minute video of the event, but the transcript is good too).

Running into Limits with Aurora? — Aurora hits walls with regional HA, failover lag, and manual sharding. CockroachDB speaks Postgres, but goes further — with global consistency, zero-downtime ops, and effortless scale. Watch the side-by-side tech breakdown in this on-demand webinar.

Autobase 2.2.0: A Self-Hosted DBaas Platform for Postgres — For when you want something akin to RDS or Cloud SQL, but on your own hardware. It automates deployment and maintenance, ensuring high availability, scalability, and cost efficiency. v2.2 adds TLS support across all cluster components, ARM compatibility, and automated backups to Hetzner Object Storage.

Debugging Postgres More Easily — Hans-Jürgen shares a technique for using table inheritance to debug transactions in an interesting way.

📊 RTABench: A Benchmark For Real Time Analytics — An open source benchmark built by the Timescale team that differs from many classical database benchmarks by focusing on real-world queries across normalized tables and incremental pre-aggregated materialized views.