KwikQuery LLC is a tech company focused on dramatically improving the performance of Apache Spark for complex analytical workloads.The flagship product, TabbyDB, is a specialized fork of Spark that tackles common bottlenecks — like long compile times, “out of memory” errors, and slow nested join queries — without requiring code rewrites or changes to existing clusters.

TabbyDB introduces advanced optimizations at both compile time and runtime. By optimizing key query-planning rules, reducing unnecessary metadata calls, and intelligently handling broadcast hash joins, TabbyDB significantly reduces the resource demands of large, deeply nested SQL workloads.

kwikquery.com

It also improves in-memory plan caching and collapses query trees early in execution, which helps avoid size blow-up during analysis.

Despite its enhancements, TabbyDB remains fully compatible with Spark’s DataFrame API, meaning organizations can adopt it with minimal disruption.

In real-world tests (e.g., on the TPC-DS benchmark), KwikQuery reports up to a 13% improvement in execution time compared to stock Spark.

Overall, KwikQuery LLC positions itself as a performance enabler for enterprises running large-scale data workloads, allowing data teams to execute complex queries faster and more reliably.