A Framework and Algorithms for Energy Efficient Container Consolidation in Cloud Data Centers

Abstract

Virtualization technology is the basic technology that causes cloud computing happening today. Virtualization means to create a logical instance of the hardware and the software. Virtualization can be carried out by traditional method or lightweight virtualization. Traditional virtualization consists of virtual machines. Lightweight virtualization technology also named as Docker getting enormous attention in past few months. Container technology emerged from the Linux world, based on key features in the Linux kernel, including cgroups and namespaces. Containers are increasingly gaining popularity and becoming one of the major deployment models in cloud environments. CloudSim framework allows to: 1. Test application services in repeatable and controllable environment. 2. Adjustment of the system bottlenecks before deploying apps in actual cloud. 3. Experiment with different workload and resource performance scenarios on simulated infrastructure for developing and testing adaptive application provisioning techniques. ContainerCloudSim which is extension of CloudSim provides support for modeling and simulation of containerized clouds. The implementation of ContainerCloudSim is divided into two main parts: simulated elements and simulated services. Researchers can utilize ContainerCloudSim to study various VM and container mapping algorithms. ContainerCloudSim is also used to look over the effect of container placement algorithms and overbooking of containers on number of container migrations, total power consumption in data center and resulting SLA violations. For this three placement policies are represented: FirstFit, MostFull and Random. These algorithms results in higher violations and energy consumption. As a proposed work by applying greedy approch on container overbooking of FirstFit algorithm for betterment in container migration rate and also energy efficiency.