Dual granularity dynamic mapping with packetized storage

A method comprising:

• organizing a data storage media into sets of atomic writing units, the sets of atomic writing units being grouped into garbage collection units;
• receiving first write data from a host, the first write data being logically located in a range of logical block addresses;
• writing the first write data to the data storage media in one or more atomic writing units;
• updating an entry of a forward map to reflect a physical location of the first write data in the one or more atomic writing units;
• receiving second write data from the host, the second write data being logically located within the range of logical block addresses;
• writing the second write data to the data storage media in one or more different atomic writing units;
• and
• updating the entry of the forward map to include a fine-grained forward map, the fine-grained forward map reflecting the physical location of the second write data in the one or more different atomic writing units.

A method comprising:

• selecting, out of a plurality of data packets wherein each data packet occupies a range of physical addresses on a data storage media, a first data packet containing at least host data having a range of logical block addresses;
• reading the host data stored in the first data packet;
• identifying one or more subsets of superseded data in the host data stored in the first data packet;
• locating one or more superseding data packets out of the plurality of data packets, the one or more superseding data packets superseding the one or more subsets of superseded data stored in the first data packet;
• selecting one or more free atomic writing units on the data storage media;
• writing valid data from a set of valid data into the one or more free atomic writing units, the set of valid data including the host data stored in the first data packet and the superseding data.