Systematic Mapping Study on Performance Scalability in Big Data on Cloud Using VM and Container
In recent years, big data and cloud computing have gained importance in IT and business. These two technologies are becoming complementing in a way that the former requires large amount of storage and computation power, which are the key enabler technolog
- PDF / 590,488 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 110 Downloads / 201 Views
Institute of Natural and Applied Sciences, Atilim University, Ankara, Turkey [email protected] 2 Faculty of Engineering, Atilim University, Ankara, Turkey {ziya.karakaya,ali.yazici}@atilim.edu.tr
Abstract. In recent years, big data and cloud computing have gained importance in IT and business. These two technologies are becoming complementing in a way that the former requires large amount of storage and computation power, which are the key enabler technologies of Big Data; the latter, cloud computing, brings the opportunity to scale ondemand computation power and provides massive quantities of storage space. Until recently, the only technique used in computation resource utilization was based on the hypervisor, which is used to create the virtual machine. Nowadays, another technique, which claims better resource utilization, called “container” is becoming popular. This technique is otherwise known as “lightweight virtualization” since it creates completely isolated virtual environments on top of underlying operating systems. The main objective of this study is to clarify the research area concerned with performance issues using VM and container in big data on cloud, and to give a direction for future research.
1
Introduction
Big data applications continue to receive an ever-increasing amount of attention, thus they become a dominant class of applications deployed over virtualized environments [1]. On the other hand, the resource utilization feature of cloud computing is mostly based on virtualization techniques, which is the common way to run different services on the cloud [2]. By combining these two, most of the big data on cloud environments are using hypervisor to provision the virtual machines. In this technique, the VMs have their own operating systems which run on the virtual hardware resources provided by hypervisor [3]. Although it is proven to be a very useful technique in resource utilization, still there is an inherent overhead because of the hypervisor [1]. In recent years, containers, which are also called “lightweight virtualization”, are gaining popularity due to their ability to offer superior performance because they do not have their own operating systems [2]. Instead, they use the OS kernel underlined with the host machine and they work similar to a regular c IFIP International Federation for Information Processing 2016 Published by Springer International Publishing Switzerland 2016. All Rights Reserved L. Iliadis and I. Maglogiannis (Eds.): AIAI 2016, IFIP AICT 475, pp. 634–641, 2016. DOI: 10.1007/978-3-319-44944-9 56
Systematic Mapping Study on Performance Scalability
635
application and are completely isolated from each other as well as from the underlying system. This technique receives its popularity mostly in Linux OS virtualization, since it uses the features provided by Linux OS kernel itself, such as “cgroup”, “namespace”, etc., in order to completely isolate each container from the rest. In this study, along with the other research questions, the main purpose of investigat
Data Loading...
