Presentation
Scalable and Self-Organizing Server Clusters for QoS-aware Applications
Computer Engineering and Systems, The 2006 International Conference on
(2006)
Abstract
We present a new scalable and decentralized design for a server cluster that provides a single service with a certain QoS objective: maximum response time guarantee. Our design is based upon providing autonomy for each node in the cluster by having the node maintain only partial information about the system and having a mechanism for each node to process jobs or redirect it to a neighbor node. Also, each node decides whether it should remain an active node in the system or got to a standby mode, based upon information about its neighbor node and its own utilization. The metrics for the system performance are the rate of rejected requests in the system, the average experienced response time for service requests, the system size, and the ability of the system to react rapidly to changes in external load or node failures. Our system performed well in all metrics compared to a similar system that provides the same goals
Keywords
- Peer to peer computing,
- Delay,
- Web and internet services,
- Web server,
- Scalability,
- Resource management,
- Switches,
- Application software,
- Design engineering,
- Job desig
Disciplines
Publication Date
Fall November 5, 2006
Citation Information
Samah A. Senbel. "Scalable and Self-Organizing Server Clusters for QoS-aware Applications" Computer Engineering and Systems, The 2006 International Conference on (2006) Available at: http://works.bepress.com/samah-senbel/11/