FAQ

How is this different from Trino?

Trino is a distributed query engine. FederIQ is a single binary built on DuckDB. For small-to-medium deployments you get most of Trino's federation story without the JVM ops burden. For true petabyte multi-tenant workloads, Trino is still the right answer.

Can I use FederIQ in production?

Today: probably not without caveats. Tier-1 features (auth, TLS, policy enforcement, observability) are shipped but have only been used by early-alpha testers. Wait for v1.0 for load-bearing workloads.

Does it modify my data?

No. Every source is attached READ_ONLY. FederIQ only reads.

What SQL dialect?

DuckDB's, which is broadly Postgres-compatible. Functions like date_trunc, epoch, list_aggregate, and DuckDB's extensions are all available.

Why is the first build slow?

DuckDB's C++ core compiles from source via the bundled feature. First build takes 3–6 minutes. Subsequent incremental builds are fast.

How do I pin data to a region?

Use the require_region field of a policy. See enforcement.

Can I cache results across processes / nodes?

Yes — use the disk cache (shared filesystem) or the Redis cache (cargo install federiq-cli --features cache-redis).

What about joins that span sources?

DuckDB plans them natively. For expensive cross-source joins consider caching the hot side or materializing to Parquet. A cost-aware planner is on the roadmap.

Is there a managed cloud?

Planned. Core will stay Apache 2.0 forever; hosted and enterprise features will be separately licensed.