مقاله انگلیسی رایگان در مورد راه حل “بدون مرکز داده” برای محاسبات ابری – IEEE 2017

The current Cloud Computing services are based on the “data center” approach, where hundreds of thousands of dedicated servers are setup to give the services. Setting up the data center for cloud is expensive and running the infrastructure needs expertise as well as a lot of resources such as high power for cooling, redundant power for assured availability, etc. For example, 45% of the data center cost goes to the acquisition of servers, 25% goes to specialized infrastructure for fault tolerance, redundant power, cooling systems, and backup batteries, while electrical cost consumed by the machines accounts for 15% of the amortized total cost [1]. In addition to the vast number of servers used in data centers, there are billions of Personal Computers (PCs) owned by individuals and organizations worldwide. These PCs are mostly underutilized, usually used only a few hours per day. Researches show that desktop computers owned by organizations are idle up to 97% of the time [2]. We had argued [3] that we shall treat the untapped CPU cycles and disk spaces of the great many underutilized PCs as precious assets, like monetary assets, to consolidate and reuse them for the good of the society and of the individuals just like the way that a credit union works. This argument had motivated an alternative Cloud Computing provision model, named “Credit Union Cloud Model” (or CUCM for short) [3]. Cloud services (mainly IaaS) built based on the CUCM are generally referred to as Credit Union Clouds (CU clouds for short). The key characteristic in CUCM is the “no data center” approach to provisioning Cloud Computing services for an institution, organization, or community. With the current public clouds, which are better called vendor clouds as they are all provided by vendors based on dedicated data centers, the concern for security/safety and loss-of-control is the primary obstacle keeping traditional IT from moving to clouds. It is understood that if the data of a business is highly confidential, the business owner is of course overly concerned about placing the data in the hands of another party. On-premise private cloud is plausibly a solution to mitigating this concern. However, the need of big upfront investment to setup the data center for the private cloud infrastructure can be prohibitively expensive. Among many other benefits of CU clouds, affordability (which means almost no additional cost for acquiring and running an on-premise cloud infrastructure) is particularly appealing. It can help an organization or business owner save up to the 45% of the cost of a data center by eliminating the upfront purchase for the cloud servers, which would otherwise be necessary. In addition, the credit union cloud infrastructure does not need additional cooling systems, which saves the additional 15% of data center’s cost on cooling. In general, our credit union cloud management system provides a feasible onpremise solution to Cloud Computing for institutions and organizations that highly care about cost and security. This paper is organized as follows. Section II reviews CUCM, our “no data center” cloud model; Section III provides an implementation overview of CUCM as demonstrated in cuCloud, and some empirical results and analysis; Section IV addresses related works; and finally, Section V concludes the paper and outlines the future work.