@supabase.com's news outlet wrote a piece about us 👀 www.thereadreplica.com/news/logical...

Caching is a bit of a cognitive monolith All of read optimization is. People don't think about it, deeply. They just default to full read replicas, or redis. We're making it better. The goal: You never have to think about your optimization layer again. www.thereadreplica.com/news/logical...

The whole way that we've been trained to think about read optimization is broken. Since PostgreSQL 10 and the introduction of Logical Replication, there's no good reason that most people running PostgreSQL at scale should have to think about read optimization

We bet early that the right way to cache Postgres was to do it in Postgres. Every hard problem we've encountered since then, that bet keeps paying us back. www.pgcache.com/blog/postgre...

Most caches are built to work with any database, then adapted to Postgres-specific features after the fact. If you're running Postgres, the database-specific features are usually the ones that matter most. PgCache is PostgreSQL first by design.

PgCache v0.5.0 + v0.6.0 shipped. Cache population and the CDC stream are two writers seeing the origin at different points on the replication timeline. We now coordinate them with LSNs so nothing stale slips in. Plus an in-memory response cache. github.com/PgCache/pgcache

PgCache PSA: We discovered a consistency issue related to query population and CDC stream coordination. We're working on it. If you've tried PgCache in the past 4-5 months and seen some issues .. we now have a good idea as to why! Cutting a new version soon with a fix 🩹

James shipped PgCache v0.4.11 a couple weeks ago. The headline change was unglamorous but matters: better query fingerprinting. WHERE x = 1 + 2 now caches the same as WHERE x = 3 Why this is more important than it sounds

no middle gear in most postgres stacks today. you either run a full read replica or you don't. nothing in between. most teams need the in-between: scale without paying to duplicate every cold table. what am i missing?

a traditional read replica copies your whole database because it has no idea what you actually read. a smart one watches the query traffic and caches what gets hit. WAL stream handles invalidation ; no TTLs, no stale window, no tuning. the shape of your traffic, not the shape of your schema.

my postgres read replica bill, itemized: data I actually query ✓ tables from 2022 I forgot exist ✓ deprecated user_sessions table ✓ 6 months of audit logs ✓ cold data that hasn't been touched since the last oncall incident ✓ you replicate all of it. every time. to serve 20% of your reads.

AWS read replicas are expensive because they copy everything to serve a small fraction of reads. PgCache is a smart read replica that caches only hot data, stays in sync via CDC, and can reduce IOPS by 40–60%. Free trial. Free small instance. Deploy in minutes.

Stop scaling Postgres with brute force. Read replicas copy 100% of your data for a fraction of queries...and add real ops overhead. PgCache: cache the hot 20% that matters. #PostgreSQL #SRE

Stop wasting money replicating 100% of your Postgres DB just to serve the hot 20% of your data. 💸 PgCache is a "Pareto Replica" proxy that uses CDC for real-time sync. No stale TTLs, no app rewrites.

Release notes for PgCache/0.4.9: 0.4.9 - 2026-04-24 Materialized query results Pre-PG18 search_path change tracking Default admission_threshold changed from 2 to 1 New mv_size_ratio setting (default 10)

Stop paying to replicate cold data. Cache the 20% that drives 80% of your traffic.

Smart DBAs are ditching read replicas and crazy IOPS bills. DM to learn more. ... or try it for yourself on our AWS AMI (v0.4.8): aws.amazon.com/marketplace/...

We are dedicated to making sure that PgDay Boston is the best it can be, and your company could be a part of that! We have sponsorship opportunities available! Talk to us today! 2026.pgdayboston.org/sponsors/ #PostgreSQL #postgres #conference

🚀 Full Stack Software Engineer @ Piq Energy – Remote (US) Shape AI-powered grid simulation! Tech: Python, FastAPI, LLM agents, Postgres, Temporal, AWS, Typescript, Svelte, GraphQL, Mapbox, D3.js Salary: $150k-$250k #FullStack #Remote #EnergyTech #AI #Hiring #Python aihackerjobs.com/company/piqe...

When you realize you could be getting more out of PostgreSQL with an automated caching proxy

If you’re scaling cloud Postgres and still adding read replicas… you’re doing it the hard way. PgCache auto-caches ~80% of queries via proxy. No app changes. No invalidation drama (CDC handles it). Some of what we support 👇 Reply "FULL" and I'll send you the full list

Scaling database teams don’t need more cache complexity. They need a way to shape database load, effortlessly. PgCache = no app changes CDC-fresh cache 80% of common query patterns covered automatically Less babysitting. More control. DM to see if we fit your stack.

Database stability with zero app changes?! (1/3) PgCache is a wire-compatible, transparent Postgres proxy that caches read data, then keeps it fresh using PostgreSQL Logical Replication Business Value 👇

Most people think CDC filtering happens after the replication stream leaves Postgres: Grab full logical stream → filter tables in app code → drop noise. ...But it's not!

“Problematic for read replicas, but perfect for caching.” That’s the best mental model for Postgres logical replication. (1/4)

We’re excited to have Google Cloud supporting Postgres Conference 2026 as our Premier Conference Sponsor. Their work in cloud infrastructure and managed PostgreSQL helps teams run at scale with reliability. Get your tickets: postgresconf.org/conferences/...

"Our query cache failed with parameterized queries because the fingerprints went wrong.” Heard that last week. It's why we're building PgCache: to support the features that matter in production, like prepared statements and partitions. Looking for 1–2 more teams to pilot (DM me)

Postgres deserves a better read cache. We're testing PgCache, our Postgres-only cache that acts like a transparent proxy. Drop-in, auto-refresh, dead simple to deploy. If this sounds useful, join the waitlist: pgache.com#waitlist