(Published on AArchitecture: Stuff June 2013)
If garbage is the forgotten by-product of mass production, today’s information economy finds its physical doppelganger in the infrastructure supporting our growing need for data storage and distribution. Stuff – understood here as the material-without-qualities of data production – takes the form of highly generic boxes arrayed around the complex security, energetic and spatial requirements of the datacentre. As the amount of digital information increases tenfold every five years and Moore’s law – which the computer industry now takes for granted – says that the storage capacity of computer chips doubles roughly every 18 months, it is evident that the physical reality of data production will become a key driver of future industrial and urban politics. While we could dismiss this scenario as essentially anti-architecture, or celebrate it as a sci-fi dystopia, it may now be the time to discuss the ways in which this phenomenon is shaping our productive landscapes. By focusing on the physicality of data production rather than on the overrated discourse on “virtual architectures”, the aim is to provide both a technical insight and a critical understanding of this increasingly pervasive reality of the built world. Browsing the web in search of information on how data is stored and secured one discovers a rapidly evolving eco-system of users, suppliers, building technologies, software, platforms, content creators, data (both BIG and small), regulatory forces, utilities, governments, energy consumption, politics, company operating tenets. (1)
Key players such as Microsoft, Google, and HP are driving the evolution of datacentres to customized modularity shifting away from the traditional image of a monolithic, raised floor factory shed full of servers towards modular, pre-manufactured components that help reduce costs, increase scalability, and improve energy efficiency. Microsoft ITPAC’s (IT Pre-Assembled Component) combine compute, power, cooling, and networking in self-contained modules allowing to right-size the initial build, respond just-in-time to business fluctuations and operate in potentially unmanned, lights-off conditions. Latest developments in open-air components technology enable the broader use of free air cooling instead of a constant flow of conditioned air drastically reducing both energy and water consumption to nearly domestic levels. These new modules supersede the traditional idea of servers in container-like boxes: today, it’s the container itself to become a computer. Information becomes matter at an architectural scale.
AOL’s Micro-datacentres are leading the way towards the miniaturization, autonomy and distribution of the storage facilities opening to a conceivable near future in which the datacentre for the post-cloud era could be small, power efficient and clustered allowing an incredible amount of geo-distributed capacity at a very low cost. This approach creates the potential for networks of micro-datacentres, each supported by solar power, that can operate together to enable a follow-the-sun strategy in which IT workloads shift across regions throughout the day. Cloud based companies will soon start to populate their own in each city because that’s more efficient than current “mega datacentres” and creates a more efficient edge network by distributing data to the edge: without the need of a physical building, Generation 4 datacenters are opening realistic scenarios of fully distributed, highly delocalized conditions of production exploiting the exponential powers of grid computing.
Current technological advancements in the industry of data storage and distribution are the bearers of the definitive transformation of an economy based on the networking of human intelligence towards a new economy based on the networking of machine intelligence: the architecture of the factory loses any traditional spatial quality to re-appear as highly generic stuff devoid of any representational content: black boxes of circuitry and silicon distributed in what remains of the old concept of the city.
In actualizing the theories of a legacy that spans from Hilberseimer, to Branzi and Natalini, this radical integration of building and information technology seems to question once again the very foundation of architecture and urban design. As the technomorphic, non-spatial and distributed character of this stuff represents the last material expression of the increasingly abstract character of our productive relationships, it may also offer a new disciplinary opportunity to re-think the production of contemporary aspects of space: the problematization of dynamic processes, the spatialization of information and the incorporation of information technologies into matter may become the necessary resources for the re-empowerment of the discipline within contemporary conditions of production.
1. “This is our first time in the situation where we couldn’t store all the information we create even if we wanted to.” John F. Gantz, The Diverse and Exploding Digital Universe. An IDC White Paper. March 2008
2. datacenterknowledge.com – A useful repository on datacentre related news
3. globalfoundationservices.com – GFS is the engine that powers Microsoft’s cloud services
4. loosebolts.wordpress.com – Michael Manos CTO of AOL Services
5. For greater details see: Philippe Morel “n extensions a’ Extensions de la grille. Sur la production contemporaine et la notation a’ partir de Le Corbusier et Ludwig Hilberseimer, in Multitudes, n.20, printemps 2005, Paris, March 2005.