Tessera

The founding team will be presenting on PgCache at Postgres Meetup for All 🐘 Wednesday, August 19 at 9AM PST / 12PM EST. - current status - how we got started, and - surprising discoveries from a year building on PostgreSQL signup here: www.meetup.com/postgres-mee...

Logical replication is for more than just ETL: building PgCache, Wed, Aug 19, 2026, 12:00 PM | Meetup

This meetup will introduce PgCache, a new Postgres project for caching hot data. Think of it as a smart read replica. The project founders James Nelson and Philip Johnston

meetup.com

2 vCPU PgCache in front of RDS, 27M rows of Stack Exchange data. read-only: origin CPU 70% to 3%. throughput 5-6x. p50 15ms to 2.5ms. If you're running a heavy read-only workload on Postgres and want to make things easier for origin ... DM me.

Our friend Kate Gawron built a white-label demand response app for a hackathon, and threw PgCache into the stack. Five grid operators, multi-tenant across four countries, and 489 unit tests is not too shabby for a hackathon build! kategawron.co.uk/2026/06/i-bu...

Kate, Cloud DBA: Expert Insights on Cloud Database Solutions

Expert Cloud DBA delivering insights on AWS, GCP, RDS & Cloud SQL for seamless cloud database management solutions.

kategawron.co.uk

PgCache doesn't just relay data to users: it keeps a conversation going with origin. I don't know of anyone else doing this for Postgres full blog post👇 www.pgcache.com/blog/postgre...

The Postgres Dividend, or How Being Postgres-First Solves Most Hard Cache Problems

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.

pgcache.com

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

Postgres' Logical Replication Stream Gets A Second Life As A Cache Invalidation Engine

Two co-founders building a Postgres caching proxy discovered that logical replication is a better invalidation mechanism than anything the caching ecosystem has come up with on its own. The hard part ...

thereadreplica.com

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

GitHub - PgCache/pgcache: PostgreSQL proxy for query aware caching, using CDC for cache maintenance.

PostgreSQL proxy for query aware caching, using CDC for cache maintenance. - PgCache/pgcache

github.com

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.

PgDay Boston @pgdayboston.bsky.social · 4mo ago

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

PgDay Boston 2026 Sponsors

We are proud to be associated with these fine sponsors, without whom PgDay Boston 2026 would not be possible!

2026.pgdayboston.org

AI Hacker Jobs @aihackerjobs.bsky.social · 4mo ago

🚀 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...