Listen to Your CPU - Full-table Scans Are Fast

QuestDB is an open-source time-series database designed for high-demand workloads, offering low latency, high ingestion throughput, and a multi-tier storage engine. It supports Parquet and SQL, ensuring data portability without vendor lock-in. Jaromir Hamala, one of QuestDB's core engineers, conducted performance tests inspired by a study comparing index merging and composite indexes, particularly focusing on QuestDB's table scanning strategy. His findings highlighted the advantages of table scanning in handling complex analytical queries, demonstrating that QuestDB could perform a full table scan of 10 million rows in under 5 milliseconds, rivaling the speed of PostgreSQL's composite indexes and outperforming MySQL's index merges. Although indexes are not deemed useless, as they can provide selectivity and performance improvements in certain scenarios, QuestDB's reliance on table scans leverages modern hardware's capabilities effectively, making it suitable for ad-hoc queries common in analytics. This approach is exemplified in a public demo where QuestDB efficiently aggregates large datasets, showcasing its focus on maximizing raw performance through optimal use of hardware.