Extreme-scale datasets
SynSQL executes queries on datasets ranging from the 1.8-billion-row Gaia DR3 catalog to 585,010-column multi-omics tables.
Predictable execution model
Queries are fast when data fits in memory. They can remain fast beyond memory when properly indexed. Full scans are always full scans — performance follows data access patterns, not heuristics.
Dense data
Traditional SQL engines are optimized for tall datasets: many rows, few columns. Wide and dense datasets become expensive because entire rows must be reconstructed before filtering. SynSQL is designed for these workloads.
SQL-first workflow
SynSQL uses standard SQL as its primary interface, including insert, update and delete statements. Users can create and drop indexes during analysis sessions to optimize repeated workloads on pivot columns. This enables workload-specific acceleration without permanently modifying the dataset. A Python client is available for workflows that require programmatic control.