Implementation and evaluation of a container management platform on Docker: Hadoop deployment as an example

In recent years, virtualization is one of the key technologies of next-generation data centers. However, the problem of virtualization technology is that each instance needs to run a client operating system and a lot of applications. Therefore, it might generate a heavy load and affect the system efficiency and performance. In this work, the performance evaluation of three environments (bare-metal, Docker containers, and virtual machines) is investigated to understand the differences between the characteristics of each environment. Also, we addressed whether container-based virtualization can solve the problems of traditional virtualization. In addition, we combined Docker with OpenStack to implement a container management platform. Finally, we took Hadoop deployment as an example to verify whether Docker can solve the deployment problem and save time.