Rebuilding Cities Right

To be sustainable, old cities need new, smarter infrastructures, says HP Labs sustainability visionary Chandrakant Patel

Since arriving at HP Labs in 1991, HP Fellow and director of HP’s Sustainable IT Ecosystem Lab Chandrakant Patel has worked to make IT systems more energy efficient. His early research in microprocessor system design led Patel to pioneer the concept of ‘smart data centers’ – data centers in which compute, power and cooling resources are provisioned based on the need. He now extends his vision of energy efficiency beyond the data center to what he calls ‘City 2.0.’

As nations look to rebuild their aging infrastructures and at the same time take on the challenge of global climate change, Patel argues that resource usage needs to be at the heart of their thinking. And, we must take a fundamental perspective in examining “available energy” in building and operating the infrastructure. Only if we use fewer resources to both build and run our infrastructures, he says, will we create cities that can thrive for generations to come. And we can only build in that way, he suggests, if we seamlessly integrate IT into the physical infrastructure to provision the resources – power, water, waste, etc - at city scale based on the need.

We sat down with Patel recently to hear more about his vision of building City 2.0, enabled by a Sustainable IT Ecosystem.

So you started out by addressing energy use in the data center?

That’s right. When we created the Thermal Technology Research Program at HP Labs in the early 90s, our industry was not addressing power and cooling in the data center at all. But we thought the data center should be looked at as a system. And if you look at it that way, there are three key components to the data center: computing, power, and cooling. We felt all of these should be provisioned based on need. Just as you dedicate the right computing instrument to the workload, you supply the power and cooling on an as-needed basis. You use sensors and controls, so that when workload comes in, you decide what kind of workload it is and give it the right level of compute, power, and cooling.

What kind of impact does this have on energy use?

Well, we built a “smart” data center in Palo Alto and a large data center in Southern India as a proof of concept. In the data center in Southern India, we used 7,500 sensors to record the temperature of its various parts, which feed back to a system that automatically controls all the air conditioners. In addition to saving 40% in energy used by the cooling system, the fine grained sensing allowed us to dynamically place workloads and shut machines down that are not being used. Furthermore, with 7500 sensors polling every few seconds, we are able to mine sensor data to detect “anomalies” so we can extend the life of large scale physical systems such as compressors in the cooling plant. This type of sensing and control is critical for large scale physical installations. One wouldn’t run a house without a thermostat, so why should one run a multi-megawatt data center without fine grained measurement and control? A ceiling fan in a house uses a few hundred watts, and it has a knob so one can change its speed based on the need. The blowers in air handling units inside a data center use10 kilowatts, and are often running at full speed all the time regardless of the data center’s needs!

How do you apply this kind of approach over the entire IT ecosystem?

First, you need to ask: what is the ecosystem? The world has billions of service-oriented client devices, like our laptops and handhelds. Then it has thousands of data centers, and thousands of print factories. That’s the ecosystem. Then you need to ask if that ecosystem is as energy efficient as it can be. To do that we take a life cycle approach. We look at the energy it takes to build and operate IT products over their life-cycle. If you do that, you can see that you might design, build and operate them in completely different ways – through appropriate choice of energy conversion means and appropriate choice of materials - ultimately leading to least energy, least material designs. Indeed, we believe that taking such an “end to end” view in design and management is required to reduce the cost of IT services that will enable the billions to use IT ecosystem to meet their needs.

Can you give an example?

Take a laptop as an example. How much energy is required to build a laptop - to extract the material, to manufacture it, operate it and ultimately reclaim it? Using Joules of available energy consumed as the currency, one can examine the supply chain and design the laptop with appropriate choice of materials to minimize the consumption of available energy. Such a technique also allows one to examine the carbon emission across a product life cycle. This type of proactive approach is good for the environment and good for business. Good for business because, in our opinion, such an approach will lead to lowest-cost products and services.

Is there an impact on IT services too?

Absolutely. Today, I can reserve train tickets online for rail travel in India from my home in the US. But most of the 700 million people in India must take a motorized rickshaw to the train station, and spend half a day, to get the ticket. They can ill afford to spend the time. Couldn't we give them appropriately priced IT services so they can do it online? That's what Web 2.0 is about for me -- meeting the fundamental needs of a society. Furthermore, these kinds of services would reduce congestion and reduce consumption of available energy. We can ask - and we need to ask - the same kinds of questions when we are talking about bringing people all kinds of resources more effectively.

How do you get the information you need to make decisions based on energy used over the life of a product?

Firstly, at design time, the IT ecosystem enables us to create a tool for analysis based on scientific principles rather than anecdotes and rules of thumb. Secondly, the IT ecosystem provides us the ability to avail energy and material data for lifecycle analysis in design phase e.g. the available energy used in extracting Aluminum from Bauxite.

Next, during operation, you use sensors and controls to manage your resources. Take traffic flow in a city. All you need to manage it is a backbone, the sensors, the data center and a panel where you can collect all that information and manage it. With that we can manage the flow so that available energy is being provisioned based on the need. You can do the same with electricity, water, waste, etc.

Thus, you are using the IT ecosystem to have a net positive impact by deconstructing conventional business models – you're creating a sustainable ecosystem using IT.

Is that what you mean by the City 2.0 ?

Yes. We started the Sustainable IT Ecosystem Lab at HP Labs because we wanted to integrate the IT ecosystem into the next generation of cities - what I've called City 2.0. If you had to build a city all over again, how would you build it? Are you going to just build a city with more roads, more bridges? Or are you going to use the IT ecosystem so that more people can use less of those physical resources more effectively? Wouldn't you think it would be better if a data center was there, and it managed all the resources? Wouldn't it be better to harvest the rain that falls in the area and have a lot of local reservoirs? Wouldn't it be good to have a local power grid instead of bringing power from somewhere else? Those are the kinds of questions that we are wrestling with.

How can HP contribute to building the City 2.0?

HP has the breadth and the depth – the billions of service-oriented client devices, the thousands of data centers and the thousands of print factories. HP covers all aspects of the IT ecosystem. And we have a great history in measurement, communication, and computation. What I’d like to see us do is leverage the past to create the future. A future where we address the fundamental needs of society by right provisioning the resources so that future generations can have the same quality of life as we do.

The US and many other countries are in recession. Building the City 2.0 is an expensive proposition, so why is it worth doing?

First of all, I think building a smart infrastructure could revitalize our economy by providing businesses with the opportunity to apply their new technologies for solving age-old problems like water distribution and energy management. And secondly, if governments around the world are going to spend on infrastructure, we probably want to do it in a smart way: not just building things for the sake of building them. We can - and should - do it in a planned, sustainable way where we also create new, high-paying and long-lasting jobs.