Scheduling Proxy: Enabling Adaptive-Grained Scheduling for Global Computing System
GRID '04 Proceedings of the 5th IEEE/ACM International Workshop on Grid Computing
Locality in structured peer-to-peer networks
Journal of Parallel and Distributed Computing
On the performance-driven load distribution for heterogeneous computational grids
Journal of Computer and System Sciences
An improved cluster-based P2P multicasting approach
Proceedings of the 2nd international conference on Scalable information systems
Fair and flexible budget-based clustering
ICC'09 Proceedings of the 2009 IEEE international conference on Communications
Towards decentralized load balancing in a computational grid environment
GPC'06 Proceedings of the First international conference on Advances in Grid and Pervasive Computing
An improved genetic algorithm for network nodes clustering
ICICA'11 Proceedings of the Second international conference on Information Computing and Applications
| Hi-index | 0.00 |
Being able to identify clusters of nearby hosts among internetclients provides very useful information for a numberof internet and p2p applications. Examples of suchapplications include web applications, request routing inpeer-to-peer overlay network, and distributed computingapplications. In this paper, we present and formulatethe internet host clustering problem. Leveraging previouswork on internet host distance measurement, we proposetwo hierarchical clustering techniques to solve this problem.The first technique is a marker based hierarchicalpartitioning approach. The second technique is based onthe well known K-means clustering algorithm. We evaluatedthese two approaches in simulation using a representativeInternet topology generated with the GT-ITM generatorfor over 1,000 hosts. Our simulation results demonstratethat our algorithmic clustering approaches effectivelyidentify clusters with arbitrary diameters. Our conclusionis that by leveraging previous work on internet hostdistance estimation, it is possible to cluster internet hoststo benefit various applications with various requirements.