A paper from Microsoft Research and the University of Virginia - The Data Furnace: Heating Up with Cloud Computing – has suggested a novel way to use the heat generated by data centres. It argues that servers could be sent to homes and office buildings and used as a primary heat source. It goes further in discussing the feasibility of Data Furnaces (DFs), small data centres of 40 to 400 CPUs that could serve as the main heat source for a single-family home.
While there’s much talk about how to reduce the energy needed to keep data centres cool, an alternative approach is to make use of the heat they generate. For example, there have been several proposals in the UK to build a data centre/business/residential complex so that the other facilities can use the data centre’s heat output.
In this study, though, the idea is to put mini data centres directly into buildings to provide cloud computing capabilities whilst also heating the building. The paper argues that by piggy-backing on only half of the energy needed for heating, the IT industry could double in size without increasing its carbon footprint or its load on the power grid.
The most appropriate locations would be office buildings
and apartment complexes. A mid-sized data centre could be hosted inside the building and the heat produced circulated using the existing heating systems. Networking and security infrastructure can be built around the data centre, with a dedicated operator to manage it.
But the paper also looked at the idea of Data Furnaces (DFs) to heat individual homes. DFs reduce the cost of conventional data centres by reducing the initial infrastructure costs, lowering operating costs and saving the expense to buy and operate home heating - the cloud service provider can sell DFs at the price of a furnace and charge household owners for home heating. The problems are the limitations of the existing power and network infrastructure, security – DFs would be in the most insecure environment – and the need for zero-touch management.
Well it’s an idea worth exploring, but there are some significant issues. For example, what happens when the buildings don’t need heat? If you also need to cool the data centre in summer then much of the gains are lost. Although I guess it could be used to heat water or for some other use instead.
In the case of home heating, what’s in it for the home owner? There will be issues of space, security, etc. that would need some incentive to make it worthwhile. The Microsoft study puts all the gains with the IT operation.
But the point really is that rather than just making data centres more energy efficient we should also be looking at the potential to make better use of the heat they generate. Unfortunately that’s not something that’s covered in the usual measures of ICT efficiency, such as the PUE, so not the basis on which data centre operators are judged. It’s another reason for a more comprehensive measure of data centre efficiency.