I’ve been playing with QuestDB and I noticed that it is impressively efficient for filtering queries (value ><= ?). Could you provide me with some intuition of why that is?
01/19/2023, 12:52 PM
There are many optimizations in QuestDB. If the filters are operating on the timestamp, then filters can be fast because they apply only to certain partitions (and rows are in timestamp order on each partition). If operating in other columns, it’s probably SIMD (parallelization) and JIT. The fact that questdb uses a columnar based model and memory mapping also explains why some operations can be quite fast. You have a lot of interesting articles at the questdb engineering blog https://questdb.io/blog/tags/engineering/