Wednesday, July 30, 2008

Green IT

Energy Saving Methods to Sustain Green IT

Abstract

The purpose of this report is to describe power saving methods of computers and data centers. Energy saving is important for both environmental and economic reasons. Idle powered computers consume significant amount of electricity. There are many power management features available in Operating Systems. Data centers consume electricity to power the servers and remove the heat which is generated by servers. Data centers can use virtualization, efficient cooling systems and smart building technologies to reduce this energy consumption. Top level IT companies plan to use green electricity for their data centers. Provide high performance computers with less power consumption is a really big technical challenge. Innovative technology has been used to face this.


Introduction

Energy saving methods have been used to create energy efficient computing. Energy efficient computing focuses on increasing the performance of computers through less power consumption. For an example single-core to dual and quad-core processors save power while increasing performance. However electricity consumption is directly effects to the environment, because carbon dioxide and other green house gases are emitted into the environment throughout the process of generating electricity. Green IT is a concept that uses to reduce the IT environmental problems and create sustainable environment.
Computer peripherals consume significant amount of electricity. Monitors, CPU and memory devices consume considerable amount of power in a PC. Likewise a laptop display consumes the highest proportion (33%) of electricity, but currently new laptops have extremely beautiful video displays (Williams & Curtis 2008, p.14). Thus optimizing the battery life of the laptops has become a big challenge for the manufacturers. Except the PCs, backend servers consume vast amount of electricity. Data centers consume electricity to power the servers and also to remove the heat generated by the servers. Data centers in US consume roughly 1.2 percent of total US electricity (Aronson 2008, p.44). Consequently data center’s energy cost is the often single largest operating cost. Currently technology is used to increase the performance of computers and servers to get high output, but at the same time business requirements are being grown faster. For an example Google was established few years ago, but now it has nearly half millions of servers (Wang 2007, p.1). That happened because lots of high performance CPUs packed into servers to accomplish the growing workload. Demand for the more computing power and functionality increase with economy expand and also with rising the living standards. Therefore computer systems has become very complex with additional features and functionalities, those consumes large amounts of electricity. Therefore electricity might become a major issue to be considered by all leading information officer in next six to twelve months (Wang 2007, p2). Thus manufactures face big technical challenge to give high performance without increasing the electricity consumption. With these circumstances rules and regulations can be created to force the use IT equipments with certain level of energy efficiency (Wang 2007, p.2). Thus energy efficient IT equipments can make significant savings for the IT companies. According to the IDC report in 1996 when IT departments around the world spent roughly 17 cents of every dollar for the electricity and 10 years later that amount has risen up to 48 cents per dollar (Maleshefski 2007, p.40). IT departments can reduce their energy consumption by following various energy management strategies. This is an individual and corporate responsibility of everyone works in IT industry.

This report reviews energy efficient strategies for personal computers for instance changing the usage and power management features in operating systems. Furthermore this review will explore the way to minimize environmental impact of data centers using virtualization, energy efficient cooling mechanisms, echo friendly data center design and green electricity. This review will also examine the new technologies which have been used to design green computers with less power consumption.

1. Energy saving methods for Computers
1.1 Usage
Lots of computers are turned on while they are not being used. The main two reasons for that are convenience and misconception. Normally computers take a few minute to be ready for use after turned it on (Weiss 2007, p.22). Thus it makes some sort of inconvenience for the users. The misconception is frequent power cycles can reduce the lifetime of computers (Weiss 2007, p.22). However lifetime of the electronic device does not depend on the frequent power cycles. It depends on electronic device’s operation time and its temperature (Murugesan 2008, p.28). Both of these factors will reduce when turn off the computers. Therefore computers must switch off while they are not being used to get efficient use of electricity and in addition to that it helps to expand the lifetime of computers.
Idle PCs consume significant amount of electricity in computer labs and also in IT companies. One factor that directly influences to this is very low individual penalty costs (Weiss 2007, p.22). That means company power bills are not paid by workers, therefore that is not a considerable issue for them. In contrast use of computers in IT companies is widespread across the whole organization. Therefore enterprise’s PC power consumption is not easily managed by IT staff (Murugesan 2008, p.28). There are several methods available to control inefficient energy consumption in the large amount of computers. Network level controller is one of them that can be used over the CPUs and monitors in computer labs to reduce the inefficient use and also network administrators can remotely awake them for maintenance (Murugesan 2008, p.28). This controller can put PC into lower power consumption mode while they are not being used.



1.2 Power Management features in operating systems
Power management features can put the computers automatically into energy saving state while they are not being used without reducing its performance. There are many types of power management features embedded in the operating systems. It is very efficient, if these features are enabled in the operating systems by default. Thus Microsoft has been enabled power management features in their latest operating system called Vista by default (Maleshefski 2007, p.42). Consequently Microsoft Vista Operating systems can automatically turn off the hard disk when the computer is idle. Although all operating system’s power management features are not deployed in default (Maleshefski 2007, p.42). Then users must find and activate the power management features.
Stand by and sleep modes are the two power management features those come with the operating systems. These power management features turn off the monitor or the displayer of computer to reduce the power consumption. This is very effective because maximum proportion of electricity is used by the displayers and the monitors, compared with the others components (Williams & Curtis 2008, p.14). Thus a considerable amount of electricity can save by turning off the displayers while computers are idle. Hibernation in windows and ‘save sleep’ in Mac are also the power management features, those can reduce the energy consumption near zero (Weiss 2007, p.22). Theses power management features can powered off all the components of the computer and this needs part of the memory to save the sates of current programs. Screen savers are also used to save energy when computers are not being used (Murugesan 2008, p.28, Weiss 2007, p.22 & Maleshefski 2007, p.42). Screen savers reduce the monitor’s energy consumption. Furthermore Black screen savers save more electricity than screen savers those display moving images (Murugesan 2008, p.28). Besides even black screen savers reduce the monitor’s power consumption by only a small proportion.




2. Energy saving methods for data centers
Data centers consume large proportion of electricity for power the servers and for the cooling mechanisms. Consequently operating cost of data centers steadily increases with the energy price increasing worldwide. The number of servers has increased by enterprises to expand their capabilities (Murugesan 2008, p.28). Since the data centers are expanded with the growth of business needs. There are several strategies to maximize the energy efficiency in data centers.
Virtualization is one popular mechanism to make data centers energy efficient. Virtualization mechanism allows deploying multiple operating systems and applications at the same time in one particular server (Maleshefski 2007, p.42). Subsequently appropriate server resources are dynamically allocated for each application without affecting the operation of applications. Furthermore virtualization can reduce electricity consumption and simplify the data center’s environment by reducing the number of servers (Murugesan 2008, p.29). Then virtualization can maximise the output of servers and minimize the inefficient use of electricity by reducing number of servers in the data centers, because virtualization can reduce number of idle physical servers that consumes electricity and generate heat.
Cooling mechanisms of data centers consume large proportion of its total electricity consumption. Data centers are mostly used air cooling systems. Air cooling system must be closely coupled with the server to save the electricity (Aronson 2008, P.44). In that case less power is consumed to push the air closer to the heat source. Liquid cooling systems are hundred times efficient than air cooling systems (Murugesan 2008, P. 29). Liquids can observe more heat compared with air. The one drawback of liquid cooling is the complexity involved with it (Murugesan 2008, P. 29). Therefore this is suitable for organizations with large high density computer infrastructures, because they can deal with these complexities and get high benefits. Furthermore IT industry invents new strategies to cooling there data centers. Subsequently IBM has started to use efficient cooling systems like fluid-cooling system and in-server, in-rack and in-row cooling systems apart from the liquid cooling systems (Murugesan 2008, P. 29). These new cooling mechanisms consume less power.
Data center design can make significant impact of its power consumption. Thus data centers must be design according to the right size (Aronson 2008, P. 44). Although some data centers have been over specified. Then lots of energy and other resources are waisted. The right size infrastructure of data centers to the workloads can cut down 50 percent of electricity bills in the real world (Maleshefski 2007, p.46). Therefore required capacity must properly identify before specifying the data center size. Furthermore there are some other aspects in data center’s layout design, those make energy efficient environment. Distributed sensors can be used in data centers to dynamically control the operations of cooling systems (Aronson 2008, P. 44). These sensors turn on the cooling systems when heat increase up to some particular level. In this situation, cooling systems are not turned on all the time. Data center layout can design with energy efficient windows and sky light (Murugesan 2008, P. 29). Then electricity is not consumed to add lights into the data centers.
Data centers can use green electricity. Green electricity means electricity that is generated using hydropower and solar panels. Environmental impact of this green electricity generation process is less. Therefore the top level companies like Microsoft and Google planned to develop new data centers using these green power sources (Weiss 2007, P. 24). This is an appropriate solution to maintain environmental sustainability with economical growth.

3. New Technologies to design green computers with less energy consumption
Computer manufacturers face difficult technical challenge to produce high performance computers without increasing the power consumption. Innovation in energy efficiency means ‘developers maintain performance levels and reduce the power consumption compared to previous design or gain more performance with same amount of energy consumption’ (Higgs 2007, p. 210). Therefore innovative technologies have been used to design green computers with less power consumption and high performance. Multicore technology is one innovative technological solution that helps to increase overall efficiency of processors with less power consumption (Higgs 2007, p.211). Likewise Dual-core and quad-core processors those come under multicore technology consume less energy while increasing performance. In earlier, number of chip operations were increased to get high performance, but this used high capacity and generated heat (Murugesan 2008, p31). Hence that is not energy efficient. New micro chip technology has used to manufacture multi-core processors with high performance and less energy consumption (Weiss 2007, p.23). This new technology makes multitasking environment within the processors. Then one processor can simultaneously do multiple tasks without increasing energy consumption.
New technology investigates about energy efficient silicon chip design. However in chip design, circuit size continuously shrinks with dramatically increase of number of transistors, but this causes to increases the power consumption and heat generation (Higgs 2007, p.211). Therefore new technology has been used to design new chip that decreases the power consumption and heat generation. New chip design focuses on reducing electricity losses and varying power consumption based on the demand (Higgs 2007, p.211). This new chip designs will help to achieve high performance with same level of power consumption.
Laptops are very popular because of convenience for use. The biggest challenge attach with the laptops is optimizing battery life with increasing performance. Thus Intel investigates on ‘Ultra Mobile’ technology, which can achieve ten times lower power consumptions than today (Higgs 2007, p.211). This technology will help to increase the performance per watt ratio. Besides a laptop displayer consumes highest proportion of electricity (Williams & Curtis 2008, p.14). Thus new laptop screen has darkened and display palette has lightened to reduce the power consumption (Murugesan 2008, p31). This method helps to optimize the battery life of laptops.
Memory of the computers also consumes significant amount of electricity. Therefore now cache memory of computers is divided into segments and each segment powered when required(Murugesan 2008, p31). This technology is very effective because whole cache is not hundred percent occupied every time. New technology further investigated that lots of energy can save by using small numbers of high capacity drivers in servers instant of using large number of small capacity drivers (Murugesan 2008, p31). Therefore less number of memory devices must maintain to achieve low power consumption.
Processor itself consumes a small amount of total electricity compared with other peripherals. Nevertheless it controls the electricity consumption of other components of the computer. Intel introduced power management mechanism called ‘Instantly available PC power management’ that puts PC into standby stage when its power consumption drops below 5 watts and it allows PC to quickly enter to the wake up stage in response to network signal (Higgs 2007, p.214). Consequently this technology is controlled by the processor and it is very convenient because PC can quickly come into ‘wake-up’ stage.
Softwares also can play considerable role in overall system power management. Power management software tool called ‘Power and thermal manager’ was introduced by Intel (Higgs 2007, p.214). This software must install into servers and PCs. Subsequently this gives advantages with built-in features like demand based switching in servers (Higgs 2007, p.214). Demand based switching reduces the inefficient power consumption of servers.


Conclusion
This review has explored that lot of idle computers consumed a considerable amount of electricity. This appears because of misconception and convenience. Hence computer users have a responsibility to reduce the inefficient energy utilization. Furthermore if users don’t like to be waiting until the computers are ready after turn it on, they can use save sleep or hibernating, because that doesn’t take much time. Apart from that users must aware about the power management facilities available in the current operating systems, because all the operating systems are not activated them in default. Then user must find and activate them.
Employees of the IT Companies are not individually penalized for inefficient energy utilization. However they have individual and corporate responsibility to reduce the inefficient electricity consumption within the organization.
Data centers must design according to the required size to reduce the electricity utilization and heat generation. Furthermore virtualization can reduce the number of servers required for the data centers. Green electricity makes less impact on the environment. Since authorized organizations must invest on new projects to generate green electricity and must motivate people to use green electricity. In future IT can make less impact on the environment by changing the user’s attitude and by the new technologies and green electricity.


Appendix

References
Aronson, J.S. 2008, ‘Making IT a positive force in Environmental change’, IEEE computer society, Vol.10 Issue 38, pp 43-45.
Higgs, T. 2007, ‘Energy Efficient Computing’, IEEE computer society, pp 210-215.
Maleshefski, T. 2007,’IT planner 5 steps to green IT’, eWeek, Vol.24 Issue 32, pp40-46.
Murugesan, S. 2008, ‘Harnessing Green IT: Principles and Practices’, IEEE computer society, Vol.10 Issue 1, pp 24-33.
Wang, D. 2007, ‘Meeting Green Computing Challenges’, IEEE computer society, Vol. 1 Issue 1, pp 1-4.
Weiss, A. 2007, ’Can the PC go green?’, ACM digital library, Vol.11 Issue 2, pp 18-25.
Williams, J. 2008, ‘Green: The new computing coat of Arms?’, IEEE computer society, Vol. 10 Issue 1, pp 12-16.

No comments: