Energy consumption and energy efficiency of computing infrastructure

Building an Optimal Digital Infrastructure is one of the four themes of our Computing for the Future of the Planet work.

Chasing renewable energy

With Prof. Andy Hopper, Dr Rip Sohan and PhD student Sherif Akoush

One of the issues to overcome with renewable energy is the problem of matching energy generation which depends on the amount of wind or solar energy available with demand which varies significantly with human behaviour. One solution is energy storage but current storage technology is limited in capacity and often far from 100% efficient. An alternative solution is to rely on geographically dispersed generation sites but this requires a large investment in power cabling to move energy around. In fact renewable generation is often at remote sites and moving power back requires a lot of expensive investment (both financially and environmentally).

We imagined the colocation of computing infrastructure with renewable generation. In times of surplus energy we could ramp up and perform more work and in times of shortage we could switch machines off stopping jobs or even migrating them to other renewable sources.

The purpose of this project was to understand the parameters of this idea. What kind of infrastructure would be required to make this feasible? How much would energy have to cost to make it desirable? What proportion of the workload would be amenable to this?

  • Sherif Akoush, Ripduman Sohan, Bogdan Roman, Andrew Rice and Andy Hopper, Activity Based Sector Synchronisation: Efficient Transfer of Disk-State For WAN Live Migration, In The 19th Annual Meeting of the IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS'11), Singapore, Jul 2011
  • Sherif Akoush, Ripduman Sohan, Andrew Rice, Andrew W. Moore and Andy Hopper, In Free Lunch: Exploiting Renewable Energy For Computing, in The 13th Workshop on Hot Topics in Operating Systems (HOTOS'XIII), NAPA, CL, USA, May 2011. [pdf]
  • Sherif Akoush, Ripduman Sohan, Andrew Rice, Andrew W. Moore and Andy Hopper. Predicting the Performance of Virtual Machine Migration. In 18th Annual Meeting of the IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS 2010), Aug 2010 [pdf]
  • Andy Hopper and Andrew Rice. Computing for the Future of the Planet. In Phil. Trans. R. Soc. A 366(1881):3685–3697, 2008. [web] [pdf]