According to the estimations the level of carbon emission in the sector of information and communication technologies triples by 2020 . This raised concern regarding the energy efficiency of the ICT devices and systems. Further we will discuss the guidelines and directives along which we can differentiate between technologies and their efficiency-level. These guidelines not only enable the customer of these services to attain a simple and easily manageable view on his consumption as technology user, but it also provides an evaluation system for local governmental and international regulators and projects. Then we examine the problem of energy efficiency in a specific field of the ICT, the server grids, as the demand for the services of this area is one of the fastest growing sections of the ICT sector. Than we rule out two possible ways of solving energy efficiency on the grids, and compare them. Finally in the conclusion we point out the main factors that make these solutions successful, and to which extent they match to our guidelines.
Table of Contents
Abstract
1. Introduction
2. Key actions and guidelines for the ICT sector
3. The challenge of the data center
4. Green Strategies of Grid: SymbioticSphere and Energy-Aware Reservation Infrastructure (EARI)
4.1 Architecture of the server grids:
4.2 Principles of behavior and the working methods
5. Conclusions
Objectives and Topics
This work examines how server grids can improve energy efficiency and compares two approaches, SymbioticSphere and the Energy-Aware Reservation Infrastructure (EARI), in relation to broader ICT energy-efficiency guidelines.
- SMART 2020 guidelines for energy efficiency
- Energy consumption and cooling in data centers
- Virtualization and shared computing resources
- Biologically inspired server-grid architecture
- Energy-aware reservation and scheduling
- Decentralized and centralized approaches to resource management
Excerpt from the Book
4.2 Principles of behavior and the working methods
SymbioticSphere has two levels of operating principles both of which are inspired by principles ruling the ecological system. The first set of these principles acts on the level of the whole system. Some of these were already mentioned, here we collect and analyze them more thoroughly.
Decentralization allows scalability of the platforms/agents letting them to adapt their needs of resources to their task requirements by avoiding single bottleneck-points in their performance and avoiding central coordination of agents and platforms. As there is no central coordination, agents and platforms are able to perform autonomously. They detect their local environment in the network, and based on this perception they interact without any intervention. Agents and platforms store and consume energy for a living. Agents perform their services to gain energy and use energy to reach network and computing resources, while platforms are gaining energy from the agents to provide them those resources and for evaporating energy periodically. Differences of energy levels trigger natural selection among platforms and agents. Scarcity of energy on an agent or platform means that there is not enough demand for its services/resources, which causes the death of these units. Contrary, higher energy level means greater demand for services or resources, which allows this agent/platform to replicate itself to increase availability.
Platforms and agents are operating under dynamic environmental conditions. Agents migrate towards platforms that are able to provide them resources that they need for their services, platforms replicate themselves on hosts with higher resource availability. Thus desired system characteristics, like adaptability, sustainability, energy efficiency and reliability emerge from these collective behaviors and interactions; they are not present in the individual agents or platforms that coexist in a symbiotic relationship, one cannot survive without another. This provides higher energy efficiency by turning off network hosts that don’t host agents.
On the unit level, platforms and agents are both carrying features of living creatures. They are able to replicate themselves in the presence of abundant energy, and they die of in energy starvation. The replicate agent gains half of the energy of its parent, and it’s based on its parent’s platform. If an agent dies the platform under it removes the agent, and releases the resources it used. The child platform also gains half of its parent energy level, but it can be hosted by neighboring hosts, in case they are healthier (see above) than the current one. Dying platforms uninstall themselves and free all the resources it has used. Agents can also migrate to neighboring platforms if this platform has more available, not yet engaged resources than the parenting platform.
Chapter Summaries
Abstract: The abstract outlines the work’s focus on ICT energy efficiency, server grids and the comparison of two approaches.
1. Introduction: The introduction sets the discussion in the context of climate change and growing ICT energy demand. It introduces the work’s guidelines and comparative focus.
2. Key actions and guidelines for the ICT sector: This chapter presents five SMART 2020 actions for measuring, monitoring and reducing energy consumption and emissions.
3. The challenge of the data center: The chapter discusses data-center energy demand, cooling and virtualization as a means of sharing computing resources.
4. Green Strategies of Grid: SymbioticSphere and Energy-Aware Reservation Infrastructure (EARI): This chapter compares the architectures and operating principles of two grid approaches, including resource allocation and the role of users.
5. Conclusions: The conclusion considers how the two approaches manage resources and relate to the SMART 2020 guidelines.
Keywords
energy efficiency, information and communication technologies, server grids, data centers, server farms, virtualization, SymbioticSphere, EARI, Grid5000, energy-aware scheduling, resource allocation, SMART 2020
Frequently Asked Questions
What is this work about?
It examines energy efficiency in server grids within the wider context of information and communication technologies.
What are its central topics?
Its topics include ICT energy guidelines, data-center cooling, virtualization, grid architecture and energy-aware scheduling.
What is the main objective?
The work aims to compare two approaches to energy efficiency in server grids and consider how they relate to ICT energy-efficiency guidelines.
What research method does the author use?
The author uses a literature-based discussion and comparative analysis of the design and operation of two grid approaches.
What does the main discussion cover?
It covers energy challenges in data centers and examines the architectures and operating principles of SymbioticSphere and EARI.
Which keywords characterize the work?
Key terms include energy efficiency, server grids, virtualization, SymbioticSphere, EARI and energy-aware scheduling.
How does SymbioticSphere draw on biology?
Its design describes agents and platforms using ecological ideas such as decentralization, natural selection, replication and symbiosis.
What role does Grid5000 play in the discussion of EARI?
Grid5000 is the platform on which the described EARI architecture is based; users reserve resources and receive energy-related feedback.
How is user participation addressed?
The comparison considers the different roles users play in the two systems, including EARI’s provision of alternative reservation schedules.
- Quote paper
- Christina Herzog (Author), 2010, Energy Efficiency of Server Grids, Munich, GRIN Verlag, https://www.grin.com/document/144464
