Handling uncertainty in cloud resource management using fuzzy Bayesian networks

Abstract

The success of cloud services depends critically on the effective management of virtualized resources. This paper aims to design and implement a decision support method to handle uncertainties in resource management from the cloud provider perspective that enables underlying complexity, automates resource provisioning and controls client-perceived quality of service. The paper includes a probabilistic decision making module that relies upon a fuzzy Bayesian network to determine the current situation status of a cloud infrastructure, including physical and virtual machines, and predicts the near future state, that will help the hypervisor to migrate or expand the VMs to reduce execution time and meet quality of service requirements. First, the framework of resource management is presented. Second, the decision making module is developed. Lastly, a series of experiments to investigate the performance of the proposed module is implemented. Experiments reveal the efficiency of the module prototype.

Measuring performance degradation of virtual machines based on the Bayesian network with hidden variables

“…Ramezani et al 28 and Shyam and Manvi 29 measured the VM performance based on massive VM properties from data center. Ramezani et al 28 and Shyam and Manvi 29 measured the VM performance based on massive VM properties from data center.…”

“…Xia et al 27 accurately predicted the states of physical machines based on hidden Markov model. Ramezani et al 28 and Shyam and Manvi 29 measured the VM performance based on massive VM properties from data center. Yan et al 30 proposed a BN-based selection algorithm to estimate the migration probability of VMs and calculate the potential total number of migrations.…”

A multi-objective optimization model for virtual machine mapping in cloud data centres

“…One of the major issues concerning both cloud service providers and consumers is real time autonomic resource management in response to highly unpredictable demands [3]. There are two main process related to optimal autonomic resource management in cloud environments: (1) virtual resource discovery and selection to execute cloud services, and (2) physical resource allocation to virtual resources -or load balancing among PMs-to cover part of the self-configuring process.…”