Abstract
Introduction
With the rapid development of smart grids, power system video data has grown explosively, creating major challenges for storage systems in terms of data volume, real-time access, load balancing, and reliability.
Methods
This study proposes a distributed storage optimization method integrating an improved CRUSH algorithm with dynamic metadata scheduling. The improved CRUSH algorithm introduces dynamic node weights and priority-based mapping rules to achieve balanced data distribution. A hierarchical distributed metadata scheduling mechanism is also developed to optimize access latency and cache hit rate.
Results
Experimental results demonstrate that the proposed method reduces data read-write latency and metadata query response time, improves node load balance and system throughput, and maintains stable performance under different load and fault conditions.
Discussion
The proposed method improves the efficiency, reliability, and scalability of distributed storage for power system video data. It provides an effective solution for meeting the storage and real-time access requirements of smart-grid video applications.