Exasol

Use it when

• Query speed on a fixed, known data volume is the goal; the in-memory columnar MPP engine tunes itself, creating and dropping its own indexes.
• Data residency governs the choice; the same engine runs as SaaS, in your own AWS, Azure, or GCP account, or on-premises.
• Predictable pricing by raw data volume beats metering compute hours.
• You want honest evaluation terms; Exasol Personal is free, unlimited in time and data, single-user.

Think twice when

• Data growth is fast or unbounded; a volume-based license prices growth at the contract line.
• You want elastic scale-to-zero economics; SaaS clusters help, but the model is closer to a sized system than a serverless one.
• Self-tuning cuts both ways: there are no hints and manual indexing is discouraged, so an optimizer miss leaves you fewer levers than Snowflake or ClickHouse offer.

How it runs

SaaS runs multiple compute clusters over shared S3-backed data; self-hosted and BYOC deployments couple in-memory columnar storage to the nodes. Indexes and statistics maintain themselves during execution. Consumer groups weight CPU and cap sessions. Iceberg and Parquet are readable natively, and virtual schemas federate external sources.