Gestión autonómica de energía y fallos en clústeres de computadores

Power consumption on big clusters is an important component of the operation cost of the data center, in particular, it increases the total cost of ownership (TCO). In this way, the implementation and usage of power optimization algorithms is crucial for an economic-wise management of the data center. The power manager introduced in this work is based on a dynamic provisioning algorithm autonomically managed that reduces human interactions and makes the system able to run using less energy and respond to errors automatically, as well. In addition, fault tolerance on big data centers is essential for the optimization of its usage and the warranty the coherence and consistency of the data al the process made on that data center. In a cluster whose nodes are dynamically provisioned, and which its topology is managed by a load balancing element, it is possible to take advantage of this feature to replace nodes with issues when free nodes are available. This work presents a fault tolerance system that integrates with the power optimization algorithm based on dynamic provisioning of nodes and enables the improvement of the fault tolerance while gua