Predição da qualidade de serviço em grades computacionais P2P.

Abstract

Peer-to-peer (P2P) grid systems have been proposed as an economical way to increase the processing capabilities of information technology (IT) infrastructures. In a P2P grid, a peer donates its idle resources to the other peers in the system, and, in exchange, can use the idle resources of other peers when its processing demand surpasses its local computing capacity. For collaborative systems like these, incentive mechanisms are needed in order to make the systems work. However, even using incentive mechanisms, the quality of the service provided by P2P grids varies significantly over time. The facts that the resources are not dedicated to the grid and that the demand is unknown bring uncertainties for QoS attributes. Despite their cost-effectiveness, scheduling of processing demands on IT infrastructures that encompass P2P desktop grids is more difficult. At the root of this difficulty is the fact that the quality of the service provided by P2P desktop grids varies significantly over time. This way, users that execute time constraint applications are compromised by this best-effort behaviour. The research we report in this work tackles the problem of estimating the quality of service of P2P grids. The models proposed are able to estimate total processing capacity that is available for a peer in the system at future periods of time. Our results show that, in general, the prediction models that uses system knowledge to perform the predictions (grey-box and white-box models) outperforms the approaches which use only historical data to apply the predictions (black-box models). In order to evaluate the prediction models, we proposed a synthetic workload generator for P2P grids.