期刊名称：International Journal of Grid and Distributed Computing
印刷版ISSN：2005-4262
出版年度：2016
卷号：9
期号：3
页码：209-220
DOI：10.14257/ijgdc.2016.9.3.22
出版社：SERSC
摘要：Cloud computing is a kind of formula mode that is based on internet, and in this way, the sharing software and hardware resource and information can be provided for the computer and the other terminal devices on demand. However, due to technical differences among main service vendors, there is a compatibility issue on appropriate format that is based on the virtual machine manager of different cloud platforms. Therefore, it brings some hinder to flexible deployment of cloud workloads to some extent. While with the great development of open source standers, such issues can be solved in another train of thought. In this paper, it proposes a constructive solution to deploy cloud workloads based on open source standers. In the whole structure, the key point is that a cloud middleware is introduced and the application of MapReduce algorithm. The cloud middleware contains four kinds of modules, namely, OVF converter, classifier, interface matched and security check. And the improved MapReduce algorithm is used to the classification phase of the cloud workloads. In the program, cloud workloads are sent to the cloud middleware, and in the process of "classified → packaged → matching interface → security check → deployment", which can make cloud workloads deployed easily and flexibly. And this paper will elaborate the progress of each section through a theory case. The biggest construction of this program is manifested in four aspects that are function modular, operation independence, security and scalability. This program must be further improvement, based on the addition of functional roles and the development of open source standards, and has a strong theoretical and practical significance.
