Currently, compaction goes over all data files and compacts them one-by-one on a single thread: given each data file is immutable and independent from each other, multiple threads could compact multiple data files at the same time, which would make the whole compaction process faster (even if IO would still be a bottleneck, so some performance numbers should be taken too).
Currently, compaction goes over all data files and compacts them one-by-one on a single thread: given each data file is immutable and independent from each other, multiple threads could compact multiple data files at the same time, which would make the whole compaction process faster (even if IO would still be a bottleneck, so some performance numbers should be taken too).