Server cluster provides high availability, scalability, and reliability by gathering server nodes into a group. Client requests need to be distributed to each server node fairly to maximize the performance of server cluster. In this paper, we propose an adaptive and efficient load balancing algorithm for the server cluster. The proposed algorithm computes the load of server nodes with the usages of computer resources and their weights. These weights are determined dynamically based on the statistics of the usages. The experimental result shows that the proposed algorithm can prevent the bottleneck of server cluster efficiently compared with existing algorithms. This guarantees its adaptability even though there are changes to the characteristic of service.
|ジャーナル||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|出版ステータス||Published - 2006|
ASJC Scopus subject areas
- コンピュータ サイエンス（全般）