Abstract
Cloud computing services that provide computing resources to users through the Internet also provide computing resources in a virtual machine form based on virtualization techniques. In general, supercomputing and grid computing have mainly been used to process large-scale jobs occurring in scientific, technical, and engineering application domains. However, services that process large-scale jobs in parallel using idle virtual machines are not provided in cloud computing at present. Generally, users do not use virtual machines anymore, or they do not use them for a long period of time, because existing cloud computing assigns all of the use rights of virtual machines to users, resulting in the low use of computing resources. This study proposes a scheme to process large-scale jobs in parallel, using idle virtual machines and increasing the resource utilization of idle virtual machines. Idle virtual machines are basically identified through specific determination criteria out of virtual machines created using OpenStack, and then they are used in computing services. This is called the idle virtual machine–resource utilization (IVM–ReU), which is proposed in this study.
Highlights
Supercomputing and grid computing have been used to process data-intensive or complex problems in parallel
The idle virtual machine–resource utilization (IVM–ReU) proposed in this study does not clean up the memory of virtual machines that are not used for a long time or migrate some computing resources to other virtual machines that require additional resources
By using the Horizon service of OpenStack, consisting of Setting, which represent basic settings such as user registration, login history, application registration and deletion, etc. to provide computing services that process a large-scale job in parallel through IVM–ReU; VM Information, which displays static, dynamic information and the states about all virtual machines created in OpenStack or idle virtual machines classified based on the idle-state determination criteria; Job Submission, by which users submit their jobs; and Result, which shows the processed task results
Summary
Supercomputing and grid computing have been used to process data-intensive or complex problems in parallel. If computing resources are provided using logically created virtual machines based on a virtualization technique, the parallel processing of large-scale jobs, which is not possible in existing systems, can be achieved, and requirements of space and hardware costs can be reduced [1,2,3,4,5,6,7,8,9,10,11,12]. When virtual machines created through virtualization techniques to overcome the limitations of existing physical servers are not used for a long period of time, the unused computing resources are wasted [13,14,15,16]. This study proposes an idle virtual machine management scheme that provides additional service functions by using idle virtual machines within OpenStack, which is a cloud computing open-source project, as a form of infrastructure as a service that supports multiple platforms
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have