At what abstraction level is the "data sharding" being done at? e.g. firmware of hard disk, file system/memory abstractsion, OS handling file/memor abstractions

At what abstraction level is the "data sharding" being done at? e.g. firmware of hard disk, file system/memory abstractsion, OS handling file/memor abstractions, application making it's own file/memory abstractions on top of OS, appliation using portable/hardware neutral data format with application going through OS for read/writes. "Why is it ...." -> File systems/memory systems are "coorinate spaces", usually accessed through OS abstractions and not directly handled by an application. Exceptions: database systems Creating a coordinate space for storage/saving at application level is equivalent to implimenting a memory allocation scheme (aka function of OS, not an application) Per scatternet link example, the "memory/storage area" for a given shard, may or maynot reside on same local machine. Application requesting "memory/storage" would not know where the "memory/storage" resides aka locally or remotely. "What am I..." -> In the context of this discussion, similar to distributed database system "Excerpt - ...." -> Sounds like a trie data structure (https://en.wikipedia.org/wiki/Trie https://en.wikipedia.org/wiki/Trie )