Presentation

Network Control in Distributed Computing for Scientific Applications

While Grid middleware allows scheduling of computational resources (CPU and storage) over many interconnected (super)computers, managing of network resources is still an open issue. Currently, the network in a grid is a shared (over provisioned) medium that offers limited services (e.g., best effort). However, the need for network resource management in grids comes from high-throughput, real-time communication in distributed applications. In this paper, we investigate network control from grid middleware in cases where network performance determines the performance of distributed applications. We present a framework in which network control is integrated into a grid middleware.

Speakers

• Mihai Lucian Cristea, University of Amsterdam

Authors

• Mihai Lucian Cristea
• Rudolf Strijkers
• Vladimir Korkhov
• Adam Belloum
• Mark Kettenis
• Aard Keimpema
• Damien Marchal
• Paola Grosso
• Cees de Laat
• Robert Meijer