Improvements of data replication algorithms In distributed storage systems

Abstract

Aim of the study. The aim of this study is to develop and improve data replication algorithms for distributed storage systems, with a particular focus on environments that employ system sharding, where each shard operates as a replicated data system consisting of multiple processors. The research seeks to enhance data availability, reliability, and performance while ensuring consistency across distributed nodes. Methodology. The proposed algorithm is designed for distributed data storage systems in which shards jointly process client requests and transactions and are capable of interacting with other shards to support inter-shard transactions. The algorithm employs distributed consensus mechanisms and coordination protocols to manage client requests, execute consensus procedures, and handle node failures. The decision to use this approach is motivated by its ability to prevent asynchronous data replication scenarios, in which multiple replicas exist but one or more of them cannot be updated or accessed by clients. The algorithm was tested using a dataset containing detailed information about an organization’s employees, enabling the evaluation of replication behavior under realistic conditions. Results. The testing process demonstrated that the proposed data replication algorithm can efficiently distribute and synchronize data across distributed nodes. The results indicate a significant increase in data access speed, a reduction in latency, and improved consistency among replicas. Additionally, the experimental findings highlight the importance of selecting appropriate replication mechanisms based on specific requirements and constraints of the distributed database environment, including data consistency, network latency, resource utilization, and fault tolerance. Scientific novelty. The scientific novelty of this study lies in the improvement of data replication algorithms for sharded distributed storage systems through the integration of distributed consensus mechanisms that effectively mitigate the problem of asynchronous replication. The proposed approach enhances coordination between replicated shards and supports reliable transaction processing across distributed segments. Practical impact. The practical significance of the study is associated with the application of the proposed replication algorithm in real-world distributed database systems. By replicating data across multiple nodes, organizations can reduce the risks related to single points of failure and improve overall system performance. The integration of advanced data replication techniques enables organizations to meet the growing demands for data availability, reliability, and performance in today’s dynamic business environment.