**LRU cache** - Requests for already decompressed data can be returned immediately instead of burning another decompression round. - Size of the LRU cache in KiB is configurable via API.