Node Load Rate Calculation Method based on Database Query Statistics in Load Balancing Strategy

Node Load Rate Calculation Method based on Database Query Statistics in Load Balancing Strategy

연구 분야: Databases

학회: 2024 Asia-Pacific Conference on Image Processing, Electronics and Computers (IPEC)

The calculation of node load rate has a crucial impact on the quality of load balancing strategies in distributed systems. The existing commonly used methods for calculating node load rates do not fully consider the current state of nodes, leading to a discrepancy between their calculation results and reality. This paper focuses on distributed information systems, taking databases as the starting point, and proposes a node load rate calculation method based on database query statistics to be used in load balancing strategies. It adopts a method of combining theoretical calculation data with benchmark performance testing data for analysis, fully considering the impact of factors such as data volume in distributed information system databases, peak database access periods, and data table indexes on calculation results. While ensuring the accuracy of node load rate calculation results, it minimizes the need for repetitive testing, conserves computer performance and energy resources, and reduces the time and economic expenses associated with node load rate calculation. The validation results of the algorithm indicate that the average error rate between the calculated database seconds per query (SPQ) and the actual values is as low as 1.7%.