You are here

Miguel Pinto defends master dissertation entitled "Distributed Hypercubes in Distributed Environments"

Miguel Filipe Dias Pinto has requested the defence of his dissertation, entitled "Distributed Hypercubes in Distributed Environments". The defence will be held in room A2 of the Departamento de Informática, on July 13, 2011, 3:30pm. The public is invited.

The examining committee is the following:

  • João M. Fernandes (UMinho) - president
  • Jorge Loureiro (IPViseu) - opponent
  • Orlando Belo (UMinho) - supervisor

Abstract

In recent years some companies have had an exponential growth of data and an increase of decision-makers consulting the Online Analytical Processing systems, thus requiring improvements in performance of these systems. The materialization and distribution of multidimensional structures across server’s nodes can help improving the system performance, without great expense. However a distributed platform is less expensive than a centralized platform, and since some organizations do not have financial capability to buy dedicated servers, using the Grid environments has become a perfect solution for the hypercube distribution. These environments can use the computational resources of the company, they do not are dedicated to data processing, and so, this platform easily gains more efficiently, without huge costs. These Grid environments can use resources geographically dispersed, heterogeneously and non-dedicated, and the equipment can be shared with partner organizations. In order to obtain a higher performance, these platforms need proper scheduling and data, so a prediction of the availability of the nodes for a moment can help this escalation, knowing which nodes have capacity to store the views, we can distribute the hypercube in advantage, improving the performance of the platform.

The application of automatic models in the distribution of views should take into account the nodes that decision makers utilize to consult multidimensional structures and the data that each one uses. These considerations in combination with the prediction of the Grid can be important for a good distribution. In this context, the distribution models will try to distribute the different views on the nodes closed to the decision makers, reducing the data transport and increasing the performance of the Grid.

05.07.2011

home contacts RSS Feed last update: 06-Dec-2019 share facebook
Drupal theme by pixeljets.com D7 ver.1.1