ICPADS '00 Proceedings of the Seventh International Conference on Parallel and Distributed Systems: Workshops
Grid resource management: state of the art and future trends
Grid resource management: state of the art and future trends
Routing, Flow, and Capacity Design in Communication and Computer Networks
Routing, Flow, and Capacity Design in Communication and Computer Networks
Overlay Networks with Linear Capacity Constraints
IEEE Transactions on Parallel and Distributed Systems
An optimal discrete rate allocation for overlay video multicasting
Computer Communications
Scalable dimensioning of resilient Lambda Grids
Future Generation Computer Systems
P2P Networking and Applications
P2P Networking and Applications
Multi-cost job routing and scheduling in Grid networks
Future Generation Computer Systems
Grid Computing: Techniques and Applications
Grid Computing: Techniques and Applications
Handbook of Peer-to-Peer Networking
Handbook of Peer-to-Peer Networking
Overlay Networks: Toward Information Networking.
Overlay Networks: Toward Information Networking.
Handbook of Metaheuristics
| Hi-index | 0.00 |
Parallel to new developments in the fields of computer networks and high performance computing, effective distributed systems have emerged to answer the growing demand to process huge amounts of data. Comparing to traditional network systems aimed mostly to send data, distributed computing systems are also focused on data processing what introduces additionally requirements in the system performance and operation. In this paper we assume that the distributed system works in an overlay mode, which enables fast, cost-effective and flexible deployment comparing to traditional network model. The objective of the design problem is to optimize task scheduling and network capacity in order to minimize the operational cost and to realize all computational projects assigned to the system. The optimization problem is formulated in the form of an ILP (Integer Linear Programming) model. Due to the problem complexity, four heuristics are proposed including evolutionary algorithms and Tabu Search algorithm. All methods are evaluated in comparison to optimal results yielded by the CPLEX solver. The best performance is obtained for the Tabu Search method that provides average results only 0.38% worse than optimal ones. Moreover, for larger problem instances with 20-minute limit of the execution time, the Tabu Search algorithm outperforms CPLEX for some cases.