Distributed Query Execution in Firebolt (w. Benjamin Wagner)

Modern analytical systems, as we see it, come in two flavors. First, you have general purpose cloud data warehouses that are very good at processing large volumes of data, but not nearly as efficient in delivering high concurrency or low-latency workloads. Second, you have accelerators like Druid and ClickHouse that are designed for high-concurrency, low-latency data serving workloads, but struggle when it comes to complex queries containing fact to fact joins or high-cardinality aggregations. At Firebolt, we aim to bridge the gap and address low-latency, high-concurrency data-intensive applications with the ease of use of general purpose data warehouses. The system we have built at Firebolt handles point lookups in milliseconds, but also scales to address complex, distributed queries.