Analysis and Optimization of Real Time Response in IoT Virtual Environment on Intel Architecture

Abstract

The Internet of Things has become popular in the past few years. Nowadays, it's growing at an immense rate in various fields. As most of the IoT applications are having constraints on resource availability due to which concept of virtualized IoT comes into the picture. In a virtual environment, OS which runs inside a virtual machine is called a guest OS. The guest OS of VM is stored on a virtual hard disk (VHD). VHD is a big, multi-gigabyte _le that is stored on the real hard disk of a computer. This virtualization process adds some delay as compared to the directly installed OS on real hardware of the system. In this work, the KVM hypervisor is used for creating a virtual machine. KVM is taken as an example as it is well known, open-source, and widely used hypervisor. The hypervisor is a type of software that communicates with actual physical hardware and allocates various resources to virtual machines. The hypervisor has its own processing requirements, which may slow down the execution of applications inside VM. Qemu-KVM creates a thread on which virtual CPU (VCPU) can run directly on a host CPU using Intel's VT-x virtualization technology. When other higher priority threads or interrupts arrive on CPU then thread switching happens. And VM exits from its running state. The virtual machine can execute its processes in between virtual machine entry and exit operations. Due to which the virtual machine gets lesser time for execution as compared to the host system. And latency increases in a virtual environment. The response time of IoT applications is relatively higher in the virtual environment as compared to a regular environment. To optimize this response time, the virtual machine requires more time to execute its applications. By modifying the system parameters, we can provide more time to the VM process for its execution. This will eventually help in the optimization of the response time of the virtual machine.