The query engine for the Iceberg lakehouse.
One binary. Embedded like DuckDB, or a server that scales out like Trino — same SQL, same Iceberg semantics, same per-query identity. Every query runs as the authenticated user, no service account, no shared root.
Build from source: cargo install --path crates/sqe-cli, Apache-2.0, Rust, no JVM.
Same query. Same result. Same identity model.
The lakehouse engine you actually own.
Scored on the public Iceberg matrix
147/162 cells across V2 and V3 — scored against the public rubric, not a vendor spec sheet.
See the matrix →Benchmark suites faster than Trino 481 at SF1
Every query differentially validated. Up to 7.2× faster at SF1, and ahead at SF10 too. Reproducible with one script.
See the receipts →Service accounts. Zero shared roots.
Every query runs as the authenticated user. The bearer token flows through to the catalog and storage. Policy is enforced before the optimizer runs.
vs Trino →Binary, laptop to cluster
A DuckDB-class embedded engine and a distributed coordinator share one SQL surface.
Cross-catalog joins, runtime ATTACH, persistent local warehouse.
Everything in one engine.
Iceberg-native read + write
Native Iceberg V2 + V3
Read and write, not read-only. Position & equality deletes, merge-on-read, copy-on-write, schema evolution.
7 catalog backends
Polaris, Nessie, Unity Catalog OSS, AWS Glue, S3 Tables, Hive Metastore, JDBC, plus Hadoop storage. Multi-cloud.
Per-query identity
Per-query identity
OIDC password grant; the user's bearer token is passed through to the catalog and S3 on every query.
Policy as plan rewriting
Row filters and column masks from Apache Ranger policies, shared with Spark and Kyuubi, enforced at the LogicalPlan layer before the optimizer runs.
Column-level lineage
Native OpenLineage 2-0-2 emitter with column-level lineage on writes. File and HTTP sinks, disk-spool fallback.
Open + everywhere
One binary, two modes
Embedded single-process engine and distributed coordinator + stateless workers, the same SQL surface, the same binary.
Three wire protocols
Arrow Flight SQL, Trino HTTP and MCP from the same coordinator. Trino clients connect unchanged; AI clients connect as the token's user, with no service account.
Rust, no JVM
Sub-second cold start. A C++/DuckDB-class footprint with a distributed story the JVM engines charge for.
Two commands to a SQL prompt.
Embedded — one binary, no cluster
git clone https://github.com/schubergphilis/sqe && cd sqe
cargo install --path crates/sqe-cli
sqe-cli --embedded # persistent warehouse at ~/.sqe/warehouse/
sqe> SELECT * FROM '/data/sales.parquet' LIMIT 5;
sqe> SELECT * FROM read_csv('s3://bucket/orders.tsv.gz');
sqe> ATTACH 'http://catalog.example.com/api/catalog' AS partner_cat
(TYPE iceberg_rest, WAREHOUSE 'analytics');
sqe> SELECT * FROM partner_cat.sales.orders LIMIT 10;Server — Polaris + Keycloak + SQE locally
git clone https://github.com/schubergphilis/sqe
cd sqe/quickstart/polaris-keycloak-client-id
docker compose up -d # Polaris + Keycloak + MinIO + SQE
# Connect over Arrow Flight SQL
docker compose exec sqe sqe-cli --port 50051 --user adminuser
sqe> SHOW CATALOGS;Full quickstart covering all seven catalog backends → docs.getsqe.com
One engine instead of two.
| SQE | Trino | DuckDB | |
|---|---|---|---|
| Embedded mode (one binary) | yes | no | yes |
| Distributed mode | yes | yes | no |
| Iceberg V2 + V3 read + write | native | V2 + partial V3 | read-only |
| Per-query OIDC passthrough | yes | service acct only | n/a |
| Ranger policy at LogicalPlan | yes | no | no |
| Multi-catalog in one engine | 7 backends | one at a time | per-extension |
| Wire protocols | Flight SQL + Trino + MCP | Trino HTTP | extension |
| Runtime | Rust, no JVM | JVM | C++ |
| Cold start | sub-second | tens of seconds | sub-second |
| OpenLineage emitter | native, column-level | plugin | no |
Performance — SF1, vs Trino 481
| Suite | SQE | Trino | Speedup | Pass |
|---|---|---|---|---|
| TPC-H (22) | 12.6s | 32.6s | 2.6x | 22/22 |
| SSB (13) | 3.2s | 5.5s | 1.7x | 13/13 |
| TPC-DS (99) | 34.2s | 95.9s | 2.8x | 98/99 |
| TPC-C (8 read) | 1.1s | 5.3s | 4.9x | 8/8 |
| TPC-E (11) | 0.7s | 4.8s | 7.2x | 11/11 |
| TPC-BB (10) | 25.4s | 96.5s | 3.8x | 10/10 |
| ClickBench (43) | 1.7s | 4.8s | 2.8x | 41/43 |
| Bank (8) | 12.0s | 37.7s | 3.1x | 8/8 |
Faster than Trino at SF1 on all eight industry suites, and on six of seven at SF10 as of the July run. Zero row differences across every run; the exceptions are shown on the
performance page, not hidden. Run your own with
BENCH_SCALE=1 ./scripts/benchmark-test.sh --compare-trino.
Recently shipped
- Oct 2026 Iceberg V3 by default
- Aug 2026 MCP, governed by the caller's own token
- Jun 2026 Read-only ops web UI