Although this doesn’t fall within my remit at work as such, it clearly is an incredibly important topic – indeed if the figures given yesterday by John Selby were accurate (computing accounts for the same energy use as the aviation industry?) then it is something we all need to be concerned about.
Starting with Peter James from www.susteit.org.uk. Outlining reasons to worry – environmental issues but also rising costs. We know there are benefits to the use of ICT, but need to look how this balances against cost. From a pilot study at the University of Sheffield, showed the following on the use of power:
- 18% of total non-residential electricity goes on IT
- 48% is on PCs
- 18% on servers
- 14% on HPC
- 10% Imaging
- 8% networking
Susteit is going to release a tool to help other institutions do these figures.
Note the high proportion on PCs – this is where we can make most significant savings (although need to look at whole lifecycle rather than necessarily just the ‘use’ phase of the lifecycle). However, the use phase is easier to deal with, and provides a clear place to take action.
What can we do? Perhaps:
- Make IT pay the energy bills (can see this for servers, but what about PCs?)
- Better, more widely applied, whole life costing models
- More cross-functional activity – especially IT and Estates
- Powerdown more; use lower power devices; use grid computing
- Complex measures like thin client; virtualisation; storage; software (invest in small footprint s/w etc.) – but do these make both environmental and business case?
Now moving to a software demo by Howard Noble from Oxford – project to get people to turn off computers when they aren’t in use. (http://projects.oucs.ox.ac.uk/lowcarbonict/about/)
Why doesn’t this happen? Possibly some myths (e.g. computers break more often if they are powercycled often) but also management issues – like applying patches – can rely on computers being on.
Howard showing a simulation that has been built that looks at energy use by PCs based on a number of different factors – work schedules, energy management policies etc. The simulation calculates energy use (and cost) so you can start to show costs. Also shows things like your capacity for Grid computing on spare CPU cycles.
The simulation gathers quite a lot of dimensions and outputs complex information to help inform decision making. Simulation tool will be free to educational institutions.
Project is going on to look at issues around validating a policy formulated based on the results of simulation, and then tools to help implement (tools like wake-on-lan functions to allow PCs to be turned off more often, but woken when needed)