PARALLEL PROCESSING IN DYNAMIC HASH STRUCTURES AND MODELING THEIR PERFORMANCE

This paper presents a new algorithm for concurrent access to extendible hashing with cautious waiting (Extendible Hashing with Cautious Waiting, EHCW), which combines two-phase locking and optimistic verification mechanisms to reduce overhead under high load. The proposed approach ensures data consistency during parallel access to dynamic hash structures without the need for full locking, addressing the challenge of efficient synchronized access under high parallelism. The algorithm enables adaptive reconfiguration of the directory structure, enhancing scalability and performance under varying loads. To evaluate the effectiveness, a simulation model of the system's operation in an environment with limited computational resources (single-core processor and disk) and a memory page buffer pool for simulating page splitting operations and dynamic directory restructuring was implemented. The developed simulation model demonstrated that the proposed algorithm outperforms traditional static hashing schemes in terms of throughput and blocking frequency, making it promising for use in operational databases, cloud computing environments, and distributed information systems, where ensuring hi