Modelización de aplicaciones auto-adaptativas para entornos de computación distribuida

Resumen

The distributed computing environments appeared for solving massive computational problems. A new paradigm known as Grid Computing emerged for solving these problems by sharing computational power and storage capacities. Moreover, the resources that compose a grid infrastructure have different geographical locations. One of the main characteristics of grid environments is the heterogeneous nature of their components. In spite of the advantages of grid computing systems, there are several problems related to tasks management, resource discovery, resource monitoring and resource selection. Considering all these problems, the adaptation concept is introduced as a feasible solution within the grid community. However, applying the adaptation at any grid level has become a challenge itself due to the grid systems characteristics and principles. This work is focused on improving the grid resource selection process with the aim of optimizing the infrastructure throughput and the application execution in this type of environments. We propose an efficient resource selection model, which identifies the resources that best fit the application requirements. The main idea is that the model provides a self-adaptive capability to grid applications. Besides, the model will be defined from the user point of view. That is to say, it does not change the grid infrastructure or its elements, it does not modify or control the behaviour of these resources, and, finally, no new policies or scheduling techniques will be applied.