Building a zero carbon Internet

[Most people still don’t appreciate the sacrifices that are going to have to be made in order to achieve CO2 reductions in line with Kyoto agreement or upcoming Copenhagen agreement as documented in a recent MIT study that public attitudes about climate change reveal a contradiction: “ Surveys show most Americans believe climate change poses serious risks but also that reductions in greenhouse gas (GHG) emissions sufficient to stabilize atmospheric GHG concentrations or net radiative forcing can be deferred until there is greater evidence that climate change is harmful. US policymakers likewise argue it is prudent to wait and see whether climate change will cause substantial economic harm before undertaking policies to reduce emissions. Such wait-and-see policies erroneously presume climate change can be reversed quickly should harm become evident, underestimating substantial delays in the climate’s response to anthropogenic forcing.”

http://climateprogress.org/2008/02/01/mit-study-sterman-stabilizing-carbon-dioxide-concentrations-not-emissions/

Overall we need 80% reduction in GHG emissions in the coming years in order to simply stabilize the amount of CO2 in the atmosphere. We have not yet begun to experience the severe climatic disruption that will occur with the existing CO2 in the atmosphere, never mind the additional GHG gasses that will be emitted in the coming years. Every sector of society is going to be affected, including the Internet and ICT industry even though their collective contribution to global GHG emissions is relatively minor at 2-3% (but projected to double about every 4 to 6 years)

Improvements in energy efficiency, or PUE ratios at data centers are simply not going to be enough, especially when you consider the overall growth in the Internet and ICT in general. The overall absolute value of GHG emissions from the Internet and ICT is bound to increase despite all our best efforts in energy efficiency. Hence, as much as possible we need to strive for a zero carbon policy. It will not be easy to move towards zero carbon in some sectors of society like transportation, that is why other sectors must make an additional effort to move to zero carbon in order to achieve an overall target of 80% reduction. (Nothing is truly zero carbon – we use the term zero carbon when CO2 emissions are insignificantly small from the energy production itself)

CANARIE in partnership with ITAC recently held a workshop on this subject on the many technical and business challenges of building a zero carbon Internet. CANARE intends to fund a pilot demonstrating the technical and business advantages of building a zero carbon Internet. Many of the presentations at the workshop are now available at:

http://www.slideshare.net/event/canarieitac-green-it-workshop

One of the challenges building a zero carbon Internet is finding renewable sources of energy to power data centers, routing nodes, carrier hotels etc. One option is to purchase renewable power from the local utility. But this is fraught with various issues in that purchasing renewable energy credits (RECs) does not eliminate the need for dirty coal plants and it does not protect the customer from significant jumps in the price of power as cap and trade comes on line. Even if the customer is purchasing RECs, renewable power delivered over the grid will increase in cost versus dirty power, as there will be significant demand for such power when cap and trade significantly pushes up the price of dirty power.

The “megawatt mindset” of most utilities also remains a major challenge to building a zero carbon Internet. Fortunately there are now companies that building windmills and run of the rivers turbines specifically for the Internet and ICT industry that will work independent of the electrical grid, even in urban settings. Windmills designed for the utility industry tend to be monstrous devices in order to get economies of scale and cause all sorts of nimby responses. But most data centers and Internet network nodes need far smaller devices that can be mounted on roofs. Since they are not part of the electrical grid, there are no transmission lines and the customer is assured of clean, renewable electricity at a long term guaranteed price.

A good example is Ecotricity in the UK which builds and operates wind turbines on partner sites such as data centers, factories and other facilities. Ecotricity takes on all the capital costs of the project, including the turbine itself. The partner agrees to purchase the electricity from the turbine and in return receive they received their own dedicated supply of green energy at significantly reduced rates. There is no finical risk to the data center or Internet node.

http://www.ecotricity.co.uk/mwp/mwp.html

A number of companies are also building specialized wind turbines for computer and telecommunication nodes that are much smaller scale in scale than the massive wind turbines used by the utilities. Ericsson for example is deploying cellular radio towers that are solely power by the wind.

http://green-broadband.blogspot.com/2008/10/wind-powered-cell-phone-tower.html

Vertical axis wind turbines originally developed at the National Research Council in Canada are also very popular. There is no rotating blades the units are extremely compact and yet can operate in all sorts of wind conditions beyond the range of normal wind turbines
http://www.windports.com/ereal.html

There is also tethered air rotor systems which provide more sustained power because they operate at higher altitudes

http://www.magenn.com/index.php
Of course moving to a less reliable energy supply imposes new challenges on the Internet architecture and routing. In the past re-booting routers or route flaps were seen as a bad thing. Manufacturers went to considerable trouble to make “carrier-class” routers and equipment that never needed re-booting. But with intermittent availability of energy at every computing and routing node we have to re-visit many of the architectural principles of the Internet in terms of topology, routing protocols, and reliability.

We need new research into such concepts as least cost Co2 path routing where the best route may not necessarily be the shortest route in terms of latency etc. New distributed caching architectures to minimize CO2 emissions such that large databases and computers can be located at distant renewable energy sites will also be required but yet do not impact latency. New network routing topologies with off-center routing and switching will be required as today most routers and switches are located at the intersection of multiple routing paths. Finally new ideas and architectures will be required as we look to integrate low CO2 optical networks with energy hungry routers.
--BSA]