A Distributed Infrastructure for Monitoring Network Resources (pp. 257-273)
Authors: (Augusto Ciuffoletti, Yari Marchetti, Department of Computer Science, University of Pisa, Pisa, Italy)
Abstract: When an infrastructure is used to process complex workflows by way of an Infras
tructure as a Service, network monitoring becomes mandatory to ensure the required
quality of service and to optimize the utilization of the whole infrastructure.
In this paper we explore this scenario, evaluate the issues that come with the net
work monitoring operation, and propose a practical solution. To support our claims,
we introduce a network monitoring infrastrucure that has been implemented as a proof
of concept for the foundations of our solution.
Keywords: Network Monitoring, Network Measurement, XML Schema Description, Grid
infrastructure, Service Oriented Architecture, CoreGRID.
The monitoring of network resources, to evaluate metrics like available bandwidth, or de
lay jitter, is a key issue in the management of distributed infrastructures. In fact, complex
workflows supported by such infrastructures may stress the performance of communication
facilities, for instance when they require the displacement of large amounts of data. To
ensure that the performance of the networking infrastructure meets the demand, the per
formance of each network element that takes an active part in the distributed computation
must be monitored before triggering a data transfer operation, or prior to engaging in a data intensive computation. A use case that is presently quite popular occurs when the Grid
paradigms are applied to an Infrastructure as a Service infrastructure.