AN INTELLIGENT REINFORCEMENT LEARNING-DRIVEN FRAMEWORK FOR PREDICTIVE RESOURCE TRACKING AND ALLOCATION IN CLOUD COMPUTING ENVIRONMENTS

Bakkala Santha Kumar, R Shankar

Abstract: Cloud computing has become a formidable paradigm of computing services that are on-demand and with scale. Nevertheless, efficient resource monitoring and allocation is an element that poses a very critical issue especially when the hardware parts that are monitored and allocated are heterogeneous and partially opaque (obscure) and the components are spread widely in extensive cloud systems. Poor visibility and poor distribution of such resources usually result to a fall in performance, high cost of operation and inefficiency in energy use. To solve these problems, the present paper suggests an intelligent framework consisting of automated hardware monitoring at the hypervisor level and predictive resource provisioning with the use of reinforcement learning. The suggested solution achieved through experimental approach is useful in providing the ability to properly manage the tracing of the elusive hardware devices even when there is a parallel event triggering and dynamic workload scenario.

Keywords: Cloud Computing, Energy-Aware Scheduling, Hardware Resource Tracking, Predictive Data Analytics, Resource Allocation, Reinforcement Learning

DOI: 10.24874/PES08.02B.002

Recieved:   Revised:   Accepted:   
UDC:

Reads: 3   

---