|
Size: 3672
Comment:
|
Size: 5639
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 5: | Line 5: |
| A single desktop computer will consume approximately £75 of electricity per year if it is left on all the time. If a computer is put into a low power state when it is not being used this cost is likely to fall to £15 per year.This figure extrapolated to the estimated 720 000 computers in UK Universities equates to an annual saving of around £40 million. The assumptions in these headline figures will not match every organisation so it is advisable to gather information about usage patterns, computer power consumption and electricity tariffs and plug them into one of the prediction models descried in this wiki. | A single desktop computer will consume approximately £75 of electricity per year if it is left on all the time. If a computer is put into a low power state when it is not being used this cost is likely to fall to £15 per year.This figure extrapolated to the estimated 720 000 computers in UK Universities equates to an annual saving of around £40 million. The assumptions in these headline figures will not match every organisation so it is advisable to gather information about usage patterns, computer power consumption and electricity tariffs and plug them into one of the [http://wiki.oucs.ox.ac.uk/ltg-public/lcict:predict_and_plan prediction models] descried in this wiki. |
| Line 7: | Line 7: |
| Switching computers into low power states when they are not needed is more problematic than turning off a light bulb or television. People familiar with the way institutional computers are used and managed will recognise at least some of the following scenarios: | Switching computers into low power states when they are not doing useful work is more problematic than turning off a light bulb or television. People familiar with the way institutional computers are used and managed will recognise at least some of the following scenarios: |
| Line 9: | Line 9: |
| * Users may need to be asked to leave their computer on for at least one day a week so that a central backup service can make a copy of files overnight | * Users may be asked to leave their computer on for at least one day a week so that a central backup service can make a copy of files overnight |
| Line 12: | Line 12: |
| * Computers are left on so that people do not have to wait for it to start up or boot | * Computers are left on overnight so that people do not have to wait for it to start up or boot in the morning |
| Line 14: | Line 14: |
| * Desktop computer is used to keep another computer on ("keep alives") * Computer is used to constantly filter emails so that email folders are organised when accessing an inbox via the web * Spare computing capacity is used by computer modelling projects e.g. SETI, climateprediction.net and GRID experiments in general |
* Desktop computers are used to keep another computer on * Computers are used to constantly filter emails so that web-based email clients are organised and more easy to use * Spare computing capacity is 'scavenged' by computer modelling projects e.g. SETI, climateprediction.net and GRID experiments in general |
| Line 18: | Line 18: |
| It is possible to implement a range of technical initiatives that can be used to allow computers to move into low power states more often. Organisations may also decide that some practices should be discouraged. | This section of the Low Carbon ICT wiki aims to describe a range of tools and practices that can be adopted by ICT managers and users to reduce the amount of energy that is wasted through the use of desktop computing infrastructures. |
| Line 22: | Line 22: |
| = Tools and techniques = == Enabling users to wake computers up remotely == It is possible to provide the ability for IT managers and end-users to wake computers that they own using another computer. By providing a facility to turn computers on remotely, it becomes possible to ask people to turn their computers to low power states more often. For instance, many organisations will ask their staff to leave their computers on so that the a backup service can access the hard drive and create a copy of files. If computers can be woken up by the backup service then users can switch them off without disrupting this essential activity. This same logic applies to other services that may need access to a computer more regularly e.g. services that install software updates, anti-virus checking and so on. |
= Tools and practices = == Waking computers remotely == It is possible to provide the ability for IT managers and end-users to wake specific computers remotely. By providing such a facility it becomes possible to switch computers into low power states more often. For instance, many organisations will ask their staff to leave their computers on so that the a backup service can access the hard drive and create a central copy. If computers can be woken up by the backup service then users can switch their machines off without disrupting this essential activity. |
| Line 26: | Line 26: |
| === Case study: University of Oxford === | (Insert a good description of WoL). If implemented successfully this approach can be used to achieve significant energy savings because users can be asked to switch their machines off at the end of each working day without effecting software update or backup schedules. The WoL service can also be used by people who need to turn their computer on when they are away from the office. Indeed such a service can be used to reduce the workload that results from computers that missed software updates and backups because they had been switched off. Unfortunately the development effort to provide a WoL infrastructure is not small and may not be appropriate for organisations that have computers distributed across a large number of subnets. === Case study: University of Oxford's WoL infrastructure === |
| Line 30: | Line 36: |
| Most computing devices are built to be able to detect when they are not being used. Desktop computers have four main power states: On (S0), sleep or standby (S3), hibernate (S4) and off (S5) which can be triggered manually, automatically after a pre-configured amount of time, and remotely via | Most computing devices are built to be able to detect when they are not being used. Desktop computers have four main power states: 1. On (S0) 1. Sleep or standby (S3) 1. Hibernate (S4) 1. Off (S5) Computers can be moved between these states manually by individuals, automatically by software installed on the computer after a pre-configured amount of time, and remotely by using a wake-on-LAN service or a service that allows a machine to be switched off remotely. It needs to be stated at the outset that many ICT managers have tried to implement automatic power management but found that: 1. Computers are put into low power states when important processes are running 1. Computers fail to go into low power states when they should 1. Computers behave unreliably after they are woken from a low power state At this stage these observations are unspecific and need further detail. If ICT managers are to feel confident that they can use the in-built power management facilities advertised by most operating system vendors then we need to better understand when these issues arise, and whether there are any solutions. === Switching computers off automatically === Case study: Switching computers off automatically at Liverpool University === Configuring standby (S3) and/ or hibernate (S4) === Case study: Configuring computers to move into standby automatically at Liverpool University === Asking users to switch their computers off === Case study: Asking users to switch their computers off at the University of Oxford Case study: Asking users to switch their computers off at the University of Leeds |
This page is under construction!
A single desktop computer will consume approximately £75 of electricity per year if it is left on all the time. If a computer is put into a low power state when it is not being used this cost is likely to fall to £15 per year.This figure extrapolated to the estimated 720 000 computers in UK Universities equates to an annual saving of around £40 million. The assumptions in these headline figures will not match every organisation so it is advisable to gather information about usage patterns, computer power consumption and electricity tariffs and plug them into one of the [http://wiki.oucs.ox.ac.uk/ltg-public/lcict:predict_and_plan prediction models] descried in this wiki.
Switching computers into low power states when they are not doing useful work is more problematic than turning off a light bulb or television. People familiar with the way institutional computers are used and managed will recognise at least some of the following scenarios:
- Users may be asked to leave their computer on for at least one day a week so that a central backup service can make a copy of files overnight
- Computers may need to be left on all the time in case there is the need to install software updates or security patches
- Users leave their computer of all the time so that they can use it remotely to for instance retrieve a file while at a conference
- Computers are left on overnight so that people do not have to wait for it to start up or boot in the morning
- Desktop computers are used as a web server or file store
- Desktop computers are used to keep another computer on
- Computers are used to constantly filter emails so that web-based email clients are organised and more easy to use
- Spare computing capacity is 'scavenged' by computer modelling projects e.g. SETI, climateprediction.net and GRID experiments in general
- Standby (S3) and hibernate (S4) power saving modes cannot be implemented reliably
This section of the Low Carbon ICT wiki aims to describe a range of tools and practices that can be adopted by ICT managers and users to reduce the amount of energy that is wasted through the use of desktop computing infrastructures.
Tools and practices
Waking computers remotely
It is possible to provide the ability for IT managers and end-users to wake specific computers remotely. By providing such a facility it becomes possible to switch computers into low power states more often. For instance, many organisations will ask their staff to leave their computers on so that the a backup service can access the hard drive and create a central copy. If computers can be woken up by the backup service then users can switch their machines off without disrupting this essential activity.
(Insert a good description of WoL).
If implemented successfully this approach can be used to achieve significant energy savings because users can be asked to switch their machines off at the end of each working day without effecting software update or backup schedules. The WoL service can also be used by people who need to turn their computer on when they are away from the office. Indeed such a service can be used to reduce the workload that results from computers that missed software updates and backups because they had been switched off.
Unfortunately the development effort to provide a WoL infrastructure is not small and may not be appropriate for organisations that have computers distributed across a large number of subnets.
Case study: University of Oxford's WoL infrastructure
[http://wiki.oucs.ox.ac.uk/ltg-public/lcict:wol_oxford_university The University of Oxford wake-on-LAN infrastructure] enables IT officers, end-users and managers of 3rd-party services to wake computers remotely.
Configuring computers to move into low power states automatically
Most computing devices are built to be able to detect when they are not being used. Desktop computers have four main power states:
- On (S0)
- Sleep or standby (S3)
- Hibernate (S4)
- Off (S5)
Computers can be moved between these states manually by individuals, automatically by software installed on the computer after a pre-configured amount of time, and remotely by using a wake-on-LAN service or a service that allows a machine to be switched off remotely. It needs to be stated at the outset that many ICT managers have tried to implement automatic power management but found that:
- Computers are put into low power states when important processes are running
- Computers fail to go into low power states when they should
- Computers behave unreliably after they are woken from a low power state
At this stage these observations are unspecific and need further detail. If ICT managers are to feel confident that they can use the in-built power management facilities advertised by most operating system vendors then we need to better understand when these issues arise, and whether there are any solutions.
Switching computers off automatically
Case study: Switching computers off automatically at Liverpool University
Configuring standby (S3) and/ or hibernate (S4)
Case study: Configuring computers to move into standby automatically at Liverpool University
Asking users to switch their computers off
Case study: Asking users to switch their computers off at the University of Oxford Case study: Asking users to switch their computers off at the University of Leeds