Sub-second Queries
Firebolt is built for performance: it has a mature planner, vectorized runtime, and fast shuffle. Fine grained control over data layout and indexing lets you run fast queries on terabytes of data. Shared computations are reused across queries.
Scaling the way you need it
Firebolt supports workload scaling by providing multidimensional elasticity. It allows you to independently and granularly scale compute resources up or down, in or out, and for concurrency, ensuring optimal performance and cost-effectiveness for your workload.
Iceberg Support
Read and write Iceberg tables. Firebolt supports file-based and REST catalogs. Queries on Iceberg run fast: our query planner, runtime, and storage are all optimized for it.
Postgres Compliant
Postgres-compliant SQL extended for modern analytics: array processing with lambdas, schema inference, and vector search.
Rich Transactions
Firebolt has decoupled metadata, storage, and compute. All clusters can read and write the same data. Transactions are ACID and offer strong consistency with snapshot isolation. Multi statement transactions and RBAC allow you to build advanced, safe applications.
Elastic Compute
Scale to zero when you don’t need to run queries. New clusters are provisioned within seconds. Auto start and auto stop allow you to optimize cost. Clusters can be re-scaled without any downtime.