Every tool you need. One platform.

Connector-based ETL supporting 300+ sources. Runs on a schedule defined by Airflow.

For sources without an Airbyte connector. Write Python scripts that run as Airflow DAG tasks.

Scheduler runs in the hot tier. Workers spin up only during task execution. Can't be disabled — orchestration is the backbone.

Primary warehouse. One Postgres instance hosts two logical databases: OrcheStack's own (platform schema) and your pipeline data (raw, marts, Airflow metadata) — kept separate for clean backups and permission scoping.

Column-store warehouse for analytical workloads above ~1 TB. Heavier than PostgreSQL.

In-process analytical engine for small datasets. No separate server; lives inside dbt.

S3-compatible object store. Open end — AI / ML / DS consumers read raw files here without going through the warehouse.

Industry-standard transformation framework. Reads from PostgreSQL raw, writes marts in-place.

Modern dbt alternative with virtual data environments and better incremental semantics.

Declarative data quality checks. Runs after every dbt build. Failed expectations block downstream tasks.

YAML-first data quality. Simpler setup; less powerful assertions than Great Expectations.

Column-level lineage across every stage. Stores data ownership, glossaries, and data contracts.

LinkedIn-origin metadata platform. Broader connector library; heavier operational burden.

Easy-to-use BI tool. Stakeholders build dashboards without writing SQL. Best for small-to-medium teams.

Advanced BI with SQL Lab, row-level security, and richer chart types. Steeper learning curve.

BI-as-code driven by dbt models. Metrics live in your dbt project; dashboards sync automatically.

Full-featured PostgreSQL web UI. Includes ERD, query plans, server monitoring. ~500 MB when running.

Lightweight single-file database UI. Only ~50 MB. Supports PostgreSQL, MySQL, and several others.

Minimal Go-based PostgreSQL client. Fastest spin-up time. Read-only query runner.