SQE mints each user's token via the OIDC password grant, then passes it through to Polaris.
Walk-through → Code ↗
Get started
Run SQE end to end.
Each quickstart is a self-contained directory that brings up the whole stack, runs real queries, and captures the output. Clone the repo, pick one, and run it — every option below is validated from a clean state.
# clone, pick a quickstart, run it end to end git clone https://github.com/schubergphilis/sqe cd sqe/quickstart/polaris-keycloak-client-id cp .env.example .env && ./run.sh
Catalog + authentication
ServerFull Iceberg catalogs with real identity, in local Docker.
AWS managed catalogs
ServerManaged AWS catalogs, bootstrapped and torn down with CDK.
Managed Iceberg; CDK bootstrap + teardown; SQE creates the namespace and round-trips.
Walk-through → Code ↗
Glue + S3; SQE creates the database and does a full create / write / read round-trip.
Walk-through → Code ↗
Lake-Formation-governed Glue: denied until an explicit LF grant, then succeeds.
Walk-through → Code ↗
Embedded
One binary, no server, DuckDB-shaped local SQL.
Read CSV / JSON / Parquet directly with the read_* table functions. No server, no catalog.
Walk-through → Code ↗
A persistent, SQLite-backed Iceberg catalog on local disk.
Walk-through → Code ↗
Attach several persistent catalogs in one session and JOIN across them.
Walk-through → Code ↗
Quack (DuckDB wire protocol)
experimentalSQE's DuckDB Quack RPC endpoint, both directions.
Walk-through → Code ↗
Operations
ServerRun SQE like you mean it.
Scrape SQE's Prometheus metrics with VictoriaMetrics and visualise them in Grafana.
Walk-through → Code ↗
Benchmarks
ServerMeasure it yourself.
Generate, load, and run the TPC suites against SQE with per-query timings.
Walk-through → Code ↗
Full walkthroughs, config, and captured output live in the docs — or browse every quickstart in the source repository.