Over the past decade, the increasing complexity of data-intensive cloud computing services along with the exponential growth of their demands in terms of computational resources and communication bandwidth presented significant challenges to be addressed by the scientific research community. Relevant concerns have specifically arisen for the massive amount of energy necessary for operating, connecting, and maintaining the thousands of data centres supporting cloud computing services, as well as for their drastic impact on the environment in terms of increased carbon footprint. This chapter provides a survey of the most popular energy-conservation and “green” technologies that can be applied at data centre and network level in order to overcome these issues. After introducing the reader to the general problem of energy consumption in cloud computing services, the authors illustrate the state-of-the-art strategies for the development of energy-efficient data centres; specifically, they discuss principles and best practices for energy-efficient data centre design focusing on hardware, power supply specifications, and cooling infrastructure. The authors further consider the problem from the perspective of the network energy consumption, analysing several approaches achieving power efficiency for access, and core networks. Additionally, they provide an insight to recent development in energy-efficient virtual machine placement and dynamic load balancing. Finally, the authors conclude the chapter by providing the reader with a novel research work for the establishment of energy-efficient lightpaths in computational grids.
TopIntroduction
Modern Information and Communication Technologies (ICTs) play an important role in people’s lives all around the world. Enormous computing capabilities in tandem with high-speed transport network allow offering a vast variety of different services ranging from media on-demand to online banking, from remote education to advanced medical breakthroughs, from online shopping to customer relationship management. These kinds of services have been commonly referred to as cloud computing services (Armbrust, et al., 2010). In literature there are various definitions of cloud computing; following (Mell & Grance, 2009), we define cloud computing as
A model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.
Cloud computing services are offered on the basis of three main models: Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS). In the first case, the cloud services are deployed in such a way that the user can exploit the computing resources offered by the cloud via machine virtualization (VM); PaaS allows the end-user running custom-built applications, eliminating the expense and the complexity associated with configuration and management of the hardware and software needed for running them. Finally, as the name suggests, the SaaS model is used for enabling access to software services on demand hosted in the cloud. The SaaS model is “built” on top of the IaaS and the PaaS models, hence processing and storage of data both happen in the cloud. Recent notable examples of this model are the Amazon Web Services (Amazon Web Services), Google Apps (Google Apps) and the Apple iCloud (Apple iCloud). Cloud computing further supports many business applications such as Customer Relationship Management (CMR), Content Management (CM), and Enterprise Resource Planning (ERP).
Unfortunately, the exploitation of the services mentioned above comes at a very big price, a price that is remarkably growing with time and that posed significant challenges for researchers and scientists all over the world. We are referring to the price associated with the massive energy demands that power-hungry data centres and network infrastructures require to respectively operate and deliver the above described cloud services to the end-user. To this day, the impact of Information and Communication Technologies (ICTs) on energy consumption has been of less concern as opposed to their computing performance until very recently and especially due to the increased price of electricity and natural gas (Europe's Energy Portal). Although computers are nowadays much more efficient compared to their predecessors of the last 10 years in terms of offered computing performance per watt (typically measured in Floating Point Operations per second per Watt or FLOPS/W), yet it can be observed that the chip power density has tremendously increased over the years. In order to deliver cloud computing services, thousands of data centres are interconnected to the backbone Internet network, hosting around 20 million servers around the world (Morgan, 2006). This introduces severe consequences in terms of total energy consumption and carbon footprint increase. To give some numbers, the power bill of the United States of America for data centre operation ranged between 2 to 3 billion dollars in 2006 (Average Retail Price of Elecrticity to Ultimate Customers: Total by End-Use Sector, 2006); it has been further estimated that in 2007 the total contribution of ICTs to the global electrical energy consumption worldwide has been of an impressive 8% and it is believed to increase further (Leisching & Pickavet, 2009).