Cloud benchmarking in bare-metal, virtualized, and containerized execution environments

Datacenters usually use virtualization methods to increase their productivity and to reduce complexity for end users while allowing resources to be utilized more effectively. Cloud computing takes the use of those resources to the next level by delivering access to those components on-demand as a service, further reducing complexity, cost, and burden. Because virtualization involves increased abstraction, there is an inevitable trade-off between these advantages and a disadvantage of potentially losing performance. As a result, there has been a considerable amount of work to develop and to improve different virtualized environments that can isolate a workload from another. Virtual machines and Linux containers are the most well-known techniques that provide an isolated environment to allow applications to run independently. Each of these methods has its own advantages and disadvantages. This thesis research aims to compare the performance of systems in bare-metal, containerized (Linux containers), and virtualized (virtual machines) execution environments. This research also analyzes and understands the implications of the benchmarking results in these different execution environments. These benchmarking results, analyses, and insights discussed can provide a guidance to cloud computing researchers and developers in the process of designing cloud computing solutions, deploying cloud systems, and developing algorithms, programs, and applications on different cloud platforms.