Performance of VM in SDN-Assisted Cloud Data Center with Working Vacations

• Main Authors: Barik, Lalbihari, Dash, Bibhuti Bhusan, Ranjan Mishra, Manoj, Rout, Suchismita, De, Utpal Chandra, Shekhar Patra, Sudhansu
• Format: Conference Proceeding
• Language: English
• Subjects: Cloud computing; Cloud Data center; Data centers; Performance Measurement; Qqueuing Model; Quality of service; Software Defined Network; System performance; Time-frequency analysis; Traffic control; Virtual machining; Working vacation
• ISSN: 2768-0673
• DOI: 10.1109/I-SMAC58438.2023.10290270

The Internet of Things (IoT)'s exponentially growing network traffic has a significant impact on energy utilization and network performance. A recommended power-saving traffic control approach for cloud data centers uses virtual machines (VM) vacation. The essential idea is that the (VM) in the cloud system may go on vacation through the utilization of software-defined network (SDN) paradigm. The reduction of active VMs while upholding QoS standards is the aim for the QoS optimization. The M/M/1 queue model with working vacations (WV) and customer requests impatience behavior is examined in this research study. This study considers the possibility that the customer's impatience is caused by a WV. The requests to be served at the cloud data center by the VMs are at slow rate than normal during WVs and are prone to get impatient. An "impatience timer" that follows exponential distribution is activated whenever customer's request reached to the cloud system and discovers that the VM is on WV. A customer leaves the queue and does not come back if their service is not finished before their timer runs out. The system's numerous performance parameters such as average system size, the sojourn time, the percentage of customer's request being served, and the abandonment frequency are calculated. Finally, some numerical findings are obtained to show how some specific parameters affect the system performance indicators.