Abstract
Distributed data storage requires swift input/output (I/O) processing features to prevent pipelines from balancing requests and responses. Unpredictable data streams and fetching intervals congest the data retrieval from distributed systems. To address this issue, in this article, a Coordinated Pipeline Caching Model (CPCM) is proposed. The proposed model distinguishes request and response pipelines for different intervals of time by reallocating them. The reallocation is performed using storage and service demand analysis; in the analysis, edge-assisted federated learning is utilized. The shared pipelining process is fetched from the connected edge devices to prevent input and output congestion. In pipeline allocation and storage management, the current data state and I/O responses are augmented by distributed edges. This prevents pipeline delays and aids storage optimization through replication mitigation. Therefore, the proposed model reduces the congestion rate (57.60%), replication ratio (59.90%), and waiting time (54.95%) and improves the response ratio (5.16%) and processing rate (74.25%) for different requests.