Storage System Dynamic Decompression for Throughput

The patent segments the data transfer process into two distinct phases: first transferring compressed data chunks between nodes, then performing decompression locally at the destination node. This segmentation allows the system to benefit from high compression ratios (reducing storage cost) while minimizing internal data transfer volume, as only the compressed data needs to be transferred rather than decompressed data. The decompression operation is segmented and executed locally, avoiding the throughput bottleneck that would occur if large decompressed data volumes were transferred internally.

The patent introduces compressed data chunks as an intermediary form between the stored data and the host-accessible data. Instead of transferring fully decompressed data across the network (which would increase network load) or transferring raw compressed data and decompressing at the host (which would increase processing complexity), the system uses compressed data chunks as an intermediate representation that can be efficiently transferred and then locally decompressed. This intermediary approach balances network throughput and processing complexity by leveraging the fact that compressed data is more compact for transfer but can be efficiently decompressed locally.