How We Increased Garbage Collection Performance with RediSearch 1.4.1

Redis introduced version 4.0 with an API for extending its capabilities via new commands and data types. RediSearch uses this feature to enable full-text search of data stored in Redis. One challenge faced by search engines is deleting and updating documents, which RediSearch tackles using a Garbage Collection (GC) mechanism. The previous approach was slow as it required scanning the entire index and acquiring a global lock. To address this issue, RediSearch developed a new GC approach that leverages Linux fork processes to minimize lock acquisition time and improve performance. This new approach is available in version 1.4.1 but is currently experimental and not recommended for production use. Future enhancements may include using heuristics to prioritize garbage collection.