WFD: Performance Optimization Strategy of Distributed Databases Cassandra

WFD: Performance Optimization Strategy of Distributed Databases Cassandra

연구 분야: Databases

학회: 2025 5th International Symposium on Computer Technology and Information Science (ISCTIS)

With the exponential growth of data, traditional centralized database faces challenges in storage and reading, and distributed databases have become a research hotspot. Apache Cassandra is a distributed NoSQL database. The Dynamic Snitch algorithm in its node selection strategy has some defects, such as the herding effect, unable to respond to the fluctuation of node performance in time, and easy to cause node synchronization problems. In order to solve these problems, this paper proposes a Weighted Feedback (WFD) algorithm, which collects key performance data such as queue length and response time in real time, dynamically scores, and adjusts scoring parameters and weights to evaluate node performance more accurately. Experiments show that the WFD algorithm is superior to the original algorithm in reducing delay and improving reading throughput under different workloads, and it also shows better performance under higher system utilization, which provides an effective solution to the problem of Cassandra replica selection strategy.