Processo de tomada de decisão para gerenciamento de alocação e migração de máquinas virtuais baseado em critérios de confiabilidade.

Abstract

Reliability is a critical factor in computer systems. In cloud environments computing reliability can be a deciding factor for machine management virtual clusters within a physical or virtual cluster. The guarantee of reliability is directly linked to factors related to the integrity of the environments in which the virtual machines are hosted. Given that an unhealthy system can compromise the entire infrastructure of a network of computers and the cluster in which it is located. The use of the TPM chip in physical or virtual form allows the aggregation of the health requirement to the virtual machines or hostesses. Allied to the verification of event logs, it is possible to use a mechanism health attestation that monitors this requirement on the nodes of a cluster. thinking Therefore, this work proposes an automatic machine allocation and migration process virtual machines within a cluster based on reliability criteria. The use of information information generated by the attestation mechanism can be used in an automated way for decision making for the allocation or migration of virtual machines within the cluster. The processes also take into account software and hardware requirements, meeting management both at the security level and at the level of verification of the resources of the cluster, making the decision of the best host for the virtual machines depending on the specified criteria and requirements. For the validation of the proposed process, a evaluation in vSphere and KVM virtualization environments. The experiments were carried out considering the software, hardware and reliability criteria and the evaluation metrics were measured from the perspective of time use and resource consumption computational. The results demonstrate that the proposed process facilitates the management allocation and migration within the cluster, in addition to optimizing the task of the manually monitor the health of created machines.