During Fortune’s Brainstorm: Green event I took part in a workshop on Sustainable IT. A key concept emerged during the session led by representatives from Microsoft, HP and Intel and moderated by Joel Makower from Green World Media: the notion of "smart infrastructure". All participants recognized the effort of the ICT industry to limit its carbon foot-print (a recent Gartner report estimates it at 2%) but the moderator reckoned that a larger opportunity might lay elsewhere. He asked the panel how they could help other industries: "What about the other 98%?"
Robert Bernard from Microsoft touched base on the role of software in sustainable IT. It is a core component to connect sensors, the grid, etc. He took the example of PG&E that announced the deployment of already 2.3 million smart meters to learn about consumption patterns, and to see what they can do with that information to reduce energy consumption from residential market (note from the Green Frog: PG&E currently provides a service to large business customers a number of energy saving programs. He stated that IT industry needed to be reinvented and that the “IP network” will become the “eP nework”. For Microsoft the interest is of course to get their operating systems in more devices like electric cars. Operating systems can help manage GPS information for instance, and identify where the next recharging station is located.
Chandrakant Patel from HP Labs presented a service point of view. A lot of progress, he noted, can be made on delivering energy services based on needs vs. the old 24x7 regulated service paradigm. He stressed the importance of looking at the carbon foot-print through the whole life cycle: design -> manufacturing -> operation -> recycling. He took the example of an Indian person who does not have electricity now and could benefit from local renewable energies for basic activities and does not need a high quality service. His group is also looking at new multi-disciplinary approach to massage all the information captured from various clouds, look at patterns and find optimization. He mentioned a lot of what was developed in the Internet bubble can be leveraged, when dot.com companies and service providers looked at consumer buying patterns.
Lorie Wigle from Intel explained that is more than chips. Obviously “smarter” infrastructure means processing and chips, and that is why they are looking closely at it. But there is a need for a system based approach, standards and open interfaces to gather and process data, and policies. She mentioned the Digital Energy Solution campaign supported by large corporations and NGOs. They are now looking how ICT can impact other industries like the automobile industry. The alliance works on smart grid, green grid projects, etc.
The Q&A session provided the opportunity to discuss practical examples. Chandrakant explained how they sample thousands of sensors every 15 seconds and how over-provisioning could be reduced with smart sensing. Robert took the example of a refrigerator that will consume twice as much as energy months before it breaks; this could be averted now if it could declare itself to the network and be monitored.
Carl Bass from AutoDesk interjected that infrastructure optimization is good but that their experience is that most of the cost and the carbon foot-print is set in the design and manufacturing phase. The CEO of AutoDesk is therefore interested to develop product libraries with green information and green design rules. He took the example of reliability commonly used today: component modules have now a lot of data about this aspect and models have been developed to create reliable products. There is no longer a need to test and crash a car to optimize its design.
Bob Bernard argued that one still needs sensors to capture the data needed to develop those models. The challenge is not much technical – most technologies are there - but behavioral. How to capture and manage all these information made possible by IT network capacities and large computing processing powers? That is the interesting part; networks will become more “human”.
The CEO of Johnston Control, also present at the workshop, concurred that a lot could be done now. A lot of infrastructure like airports has already access to a lot of information from sensors. Yet, not much has been done with it. He took the example the Empire State Building, which they helped retrofit with control systems. The building is now using 30% less energy and the upgrade cost will pay for itself in 3 years.
Joel Makower concluded the workshop by stating that there is an enormous opportunity to rebuild the infrastructure in a better way. The panel highlighted a number of areas but came short of defining what a smart infrastructure is. I will address this question in my next post before digging into the historical significance of the move towards a smart infrastructure.
Comments