Event Date : 24-07-2024
Time : 3:30pm- 4:30pm
Venue : 301, SICSR
Event Co-coordinator(s) : Ms. Priya Ambede
Speaker: Dr. Parag R. Kaveri, Deputy Director, SICSR, Pune
Event details:Cloud computing offers powerful computational services by processing tasks on virtual machines (VMs) using resource-scheduling algorithms. However, existing algorithms often yield suboptimal results due to ineffective resource scheduling. Moreover, they struggle to handle tasks that generate faults during computation, primarily due to the lack of an intelligent mechanism.
To address these issues, an algorithm called Reinforcement Learning-Shortest Job First (RL-SJF) has been developed by integrating reinforcement learning with the traditional Shortest Job First (SJF) algorithm.
An experiment conducted on a simulation platform compared the performance of RL-SJF against SJF under various scenarios involving challenging tasks. The results demonstrated that the RL-SJF algorithm improved the resource-scheduling process, achieving a 14.88% reduction in aggregate costs compared to SJF. Furthermore, the RL-SJF algorithm provided enhanced fault tolerance, successfully processing 55.52% of tasks, compared to just 11.11% for the SJF algorithm.
These findings highlight that the RL-SJF algorithm not only boosts overall cloud performance but also ensures a more reliable quality of service (QoS).
PHOTOGRAPH-
CAPTION- Discussion on Fault Tolerance in Cloud Computing
