This development adds Compaction-Aware Zone Allocation Algorithm to ZenFS.
CAZA was designed based on how the compaction process selects, merges and invalidates SSTables in the LSM tree. As such, CAZA’s design considering the compaction process of LSM-tree maximizes the zone cleaning efficiency of ZenFS by consolidating SSTables with overlapping key ranges located at different levels in the LSM-tree in the same zone and invalidating them together during zone cleaning.
Please refer Compaction-aware zone allocation for LSM based key-value store on ZNS SSDs.