Surviving climate warming - 04 Feb 2011

I continue to be amazed at how little attention is being paid by network engineers on adapting to the severe challenges of global warming. Most people think that major climatic effects from global warming such as rising sea levels and major droughts will not happen for at least another few decades. But in fact the major snow and rainfall patterns we are seeing are this year are the first significant evidence of global warming. We have seen the damage done in Queensland Australia which took out AARnet and many commercial networks. Climate models predict we are going to see a lot more of these severe events. Last year every continent on the planet suffered record breaking floods. And things are only going to get a lot worse and a lot more frequent in the coming years. For a more detailed analysis of coming threats and frequency of severe precipitation events please see

The current thinking is that these unusual snow and rainfall events are once in a lifetime occurrences, whereas in fact they are rapidly becoming the norm. The Netherlands is particularly susceptible which has suffered many massive floods in 1953 and 1995. But major flooding and storm damage is predicted for many parts of the world. If massive floods are only once a hundred year event, you can largely ignore there impact on the design of your network. But when massive destructive floods start to be a regular seasonal event new networking architectures are required.

What happens when all electrical transformer substations are flooded and out of action? How do you build a survivable network when fuel trucks can't deliver fuel to backup diesel generators because of flooded roads? How will you power network equipment in the event of such an emergency - at a time when network access is more critical than ever? R&E networks in particular may be called upon to play a critical role to provide access to HPC facilities for localized flood and climate models, as well collecting real time data from a host of sensors to measure climate and flood conditions. Schools and universities, as well, will want to continue to provide education and research services, as is practically possible, during such climatic events.

Adapting networks to survive global warming does not require significant new investments but more of a change in the mindset in the assumption that power will always be available either from the grid or standby generators. A lot of current work being doing with virtual computing, optical exchanges, federated networking, wireless services are readily adaptable to be used to build survivable networks.

In no order of priority here are some issues that network engineers need to start to address to build networks to adapt to global warming:

1. Move equipment room containing cable termination equipment, switch gear, and any other electronic equipment out of basements. I recommend moving all this equipment as close as possible to the roof where solar panels and micro windmills can be located. Optical fibers should not terminate in cable vault in basement, but instead be fused with splices that bring the fiber up to the equipment room

2. Power all critical equipment with solar panels and or micro windmills using PoE or 400 Hz systems. Design overlapping access nodes and services to provide multiple layer coverage e.g. WiFi overlaid with GSM/LTE overlaid with WhiteFi/mesh radios. If R&E network operates a MVNO they can provide these overlay networks and balance loads between GSM/WiFi etc with local handoff. See

3. Deploy solar or windmill powered VMs for critical caching and content distribution that can be rapidly relocated to other sites depending on the availability of power – see for more details. Build clouds of federated VMs to share resources see EU Contrail project -

4. Deploy OpenFlow and/or Mantychore to build virtual routers or federated forwarding tables to distribute IP routing to distant sites that may have enough power to support IP routing.

5. Build open optical exchanges and use OpenFlow with UCLP or OpenDrac to set up lightpaths to enable dynamic creation of VM clouds and distributed routing. Optical lighpaths will allow engineers to change IP topology to take advantage of available power for edge and core routing. For example, during severe floods power hungry edge routing can be moved off campus virtually using Mantychore and then looped back to campus with optical lighpaths

6. Federated open optical exchanges allow networks to exchange lightpaths between independent networks and establish new topologies to meet demand

Powering optical and routing equipment with electricity from renewable sources such as windmills and solar panels probably poses the biggest challenge. Most vendors have yet to address these issues as they are still largely fixated on the dead end thinking of energy efficiency. Educating equipment vendors to provide appropriate solutions will be probably the most important step. But deploying such a network will not only prepare the R&E community for the inevitable coming environmental disaster, it also allow significant cost savings in energy consumption and carbon emissions.

Smart Grids Home page