11-09-2014, 03:00 PM
Smart Metering the Clouds
Smart Metering the Clouds.doc (Size: 202 KB / Downloads: 8)
Abstract
As cloud computing becomes increasingly pervasive, the data center energy consumption attributable to cloud computing is climbing, despite the clarion call of action to reduce consumption and reverse environmental effects. At the same time, the rising cost of energy — due to regulatory measures enforcing a “true cost” of energy coupled with finite natural resources rapidly diminishing, resulting in scarcity — is refocusing IT leaders on efficiency and total cost of ownership (TCO), particularly in the context of the world- wide financial crisis. We propose a “smart metering” approach that encompasses all the stakeholders in the cloud computing ecosystem to achieve these twin goals of “energy conservation” and “demand response”. As such, this paper introduces our initial thoughts on “smart metering” and various implications and implementation ideas related to it.
INTRODUCTION
Cloud computing fulfills the long-held dream of computing as a utility [1]and thus represents an inflection point in the geography of computation and IT services delivery. This paradigm marks a fundamental yet massive shift from the traditional “desktop-as-a-platform” to “internet-as-a-platform” model. To achieve the infinite scalability, guaranteed performance and nearly “always-on” availability demands, these computing platforms are typically deployed in clusters of massive number of servers hosted in dedicated data centers. Each data center houses a large number of heterogeneous components for computing, storage, and networking, together with an infrastructure to distribute power and provide cooling. As the demand for cloud-based services drastically increasing in recent times, the energy consumption attributable to these services by data centers has also been skyrocketing. Recent reports also highlight this growing concern with data center energy consumption and show how current trends could make energy the dominant factor in the Total Cost of Ownership (TCO) [7]. For example, the approximately 6000 data centers in the United States consumed roughly 61 billion kilowatt-hours (kWh) of energy or about 1.5 percent of the total US electricity consumption in 2006, according to an EPA report [2]. Estimates also indicate that by 2011, data center energy consumption could nearly double [5]. Figure1 shows how power is becoming the new limiting factor in
data center costs
Cloud Data Center Costs
The rest of the paper is organized as follows. In Section 2, we discuss the various concepts, definitions and trends underlying the cloud computing paradigm, data center energy efficiency and smart metering. Section 3 explains a high level conceptual description of our vision model encompassing the various ecosystem partners and stakeholders, possible implementation methods, the key benefits to be realized and lists various pricing mechanisms to monetize our model. We finally summarize and lay future directions in Section 4 before concluding
CONCEPTS AND TRENDS
The emergence of embarrassingly distributed cloud-based applications and services brings plenty of opportunities to improve cost, scale, reliability and performance. We look at some of these trends, and try to define and understand related concepts in this section.
Emergence of Cloud and Cloud based Services
The emergence of cloud computing represents a fundamental yet massive shift in “computing” as we have known [23]. In other words, it is the next natural step in the evolution of computing in general, and IT services delivery in particular. The key idea of this paradigm is to provide a utility service, similar to a power grid, into which a user may plug-in regardless of location to access centrally located services. The metaphor is of an “electronic cloud” that follows the users as they change their physical locale.
The term “Cloud Computing” refers to an on-line delivery and consumption model for business and customer services, ranging from IT services like Software-as-a-Service (SaaS),
Storage or Server capacity as a service, and many “non-IT” services which are not “computing” tasks per se. All these
Environmental Implications
Apart from these technological implications, the growing environmental awareness in government and corporate circles on the impact of data center energy consumption is demanding for more proactive steps in this direction. Now, corporate executives are allocating time, energy and money to invest in environmental initiatives. Governments are allocating research and regulations, and drafting laws to address the efficiency of data centers and other critical components of IT infrastructure. Consumer advocates, policy makers and influential leaders are prompting IT organizations to significantly reduce the impact that computing and electronics make on the environment. This has led in creating environmental impact measurements such as carbon footprint, which is usually measured in tCO2eq (metrics tons of CO2 equivalent) based on the source of energy and amount consumed, manufacturing and logistics impact, as well as end-of-life impact (e-waste, environmental externalities etc) [6]. Most of the IT organizations are also measuring the ongoing energy costs and environmental impact of specific applications within the enterprise through the use of a set of Energy Usage Profiles (EUPs) [13] [17][19].
New Architectural Decision Points
The need to reduce power consumption is thus obvious. As we have seen, energy consumption control is a much more complex architecture problem and thus requires a coordinated array of tactics, from capacity planning techniques to optimizing operational processes and facilities design [16]. This encompasses an ongoing and holistic plan of action for developing an environmentally sustainable solution that leverages - technological advances, process frameworks and strategic initiatives across the computational ecosystem, ranging from energy suppliers, data centers, and governance bodies to end-customers. Some of the attributes of such plan can include: encouraging IT reuse, reducing IT complexity, aligning all the stakeholders,
Smart Grid
In its most basic form, a smart grid refers to an effort to prod users (consumers) of electricity to change their behavior around variable electric rates or to pay vastly increased rates for the privilege of reliable electrical service during peak conditions, and reduced rates during off-peak conditions. The grid is designed so as to reduce demand during peak usage periods by making use of the technology advances in metering and communication protocols. In this way, it relies on the free-market capitalism ideals to level the load curve assuming that the customer will always act in response to the market signals.
Smart Metering
In our definitional framework, “smart meters” represent the next generation utility measurement devices with real-time sensors, power outage notification, and power quality monitoring and come only as part of the smart grid environment. Smart meters provide a economical way of measuring “when” and “where” of energy consumption, allowing price setting agencies (energy suppliers) to introduce differential pricing based on the time of day and the season.
OUR MODEL: SMART METERING THE CLOUDS
In the previous sections, we have seen how rising fuel prices and capacity costs together with shrinking reserve margins and green house gas emissions are setting the policy maker’s agenda today. The governance and compliance regulations are placing a renewed emphasis on the demand side of the equation to deal with these issues. We believe that both the energy efficiency and demand response plays a
Pricing Model
Pricing is important in our extensible architecture, as one of the goals of “smart-metering the clouds” is to take advantage of the differentiated pricing structures available from the energy suppliers side to improve the scalability requirements and to deal with the bursts in resource demands on data centers side. Charging variable prices can be useful for resource management as it can result in the diversion of demand from high-demand time periods to low-demand time periods[32] [36].
However, this poses several challenges. At present, service providers have inflexible pricing, generally limited to flat restricted to offerings from a single provider at a time. Charging fixed prices is also not fair to both the parties since different consumers have distinctive usage scenarios and demand specific QoS for various resource requests that can change any time [33]. In addition, the scenario becomes eve more delicate with many providers having proprietary interfaces to their services, restricting the ability of customers to swap one provider for another
SUMMARY AND FUTURE WORK
Given the scale, cost and importance of emerging cloud service data centers, it is incumbent on us to rethink the way metering is done from a multi-faceted view point. We’ve shown how smart-metering can be part of a holistic & integrated solution encompassing the whole spectrum of players including energy suppliers, data centers, cloud service providers, third-party software vendors, and even
end -customers. In this way, smart-metering can be seen as the main information gathering device for the operation of a cloud utility services model and a missing link in the understanding of how energy is consumed. The other advantages include self-healing using real-time information and automatic controls to anticipate, detect and respond to system problems.