pg Koldstore

Early development — not production-ready

KoldStore is an open-source PostgreSQL extension for forever-growing app tables. It keeps the active working set in the PostgreSQL heap, moves older rows to compressed Parquet you control, and still lets you query one table with normal SQL.

This is different from pg_kalam: pg_kalam is a foreign-data wrapper that talks to a KalamDB server. KoldStore runs inside your PostgreSQL instance and tiers storage for local heap tables.

What you get

Tier	Where it lives	Optimized for
Hot	PostgreSQL heap + native indexes	Active data, low-latency reads, transactional writes
Cold	Open Parquet (filesystem, S3/MinIO, GCS, Azure)	History, cost, retention

Placement is policy-driven (hot_row_limit or age via move_after), not access-frequency measurement. Reads go through KoldMergeScan so the newest visible row wins across both tiers. Catch-up clients use koldstore.changes_since over an exclusive seq cursor.

Good fit today

Messages and chat history
AI memory / conversation transcripts
Audit logs and append-heavy events
Notifications, activity feeds, IoT telemetry

Not a good fit yet

Payment ledgers that need global uniqueness across hot and cold
Workloads that need cold point lookups as fast as a B-tree
Schema evolution, compaction, and backup/restore that are still hardening

Read this section in order

Installation — preload, Docker, from source
Getting Started — first managed table
Under the Hood — WAL capture, mirror, merge scan
Manage & Flush — policies and jobs
Change Feed — changes_since / last_rows
SQL API Reference — every shipped function and GUC
Operations — auto-flush, mirror health
Benchmarks — storage and VACUUM wins
Status & Limits — honest beta boundaries

Source and packaging

Repository: github.com/kalamdb/koldstore
License: Apache 2.0
PostgreSQL: 15–18 (published Docker image targets PG 16)
Product page: /koldstore