Are data centres becoming the power stations of the 21st Century... just a utility into which we plug our various computing devices?
Over the past year we have seen some significant mergers and acquisitions among data centre operators. Equinix bought Telecity, Digital Realty bought Telx, NTT bought e-shelter and a raft of smaller operators either bought, or were bought by hosting and services companies. On the surface this looks like the consolidation of a maturing market. Yet the market for data centre space, power and connectivity is still growing strongly and, based on the number of requests I receive to brief financial analysts about the state of the market, there is clearly a view that this is a dynamic market of some importance.
So what has changed? Not too much if you look at the data centres themselves. Issues around location, power availability, security, sustainability, operational effectiveness and the solutions to those issues remain fairly constant. The real change is going on inside and between the data centres. Here, the mega-scale internet players like Amazon, Google and Facebook have taken a forensic microscope to the way in which their computing services are provided and kick started a revolution using software abstraction and composable, industry standard infrastructure to drive a fundamentally cheaper and more flexible computer environment.
This combination of cost reduction and increased flexibility can be evidenced in where these companies locate data centres. The fact that Facebook has one of its data centres in Lulea in Sweden is partly about cost. It is cheaper to power and cool a data centre there, but because of the flexibility of their infrastructure it means that they can easily locate all our less used photos up there. The bitcoin miners that require huge amounts of power and cooling also tend to choose facilities in cool climates with access to unrestricted low cost energy. That of course begs the question about connectivity and latency.
The major carrier neutral data centres have increasingly based their market offers around the benefits of global connectivity, low latency and proximity to key markets. Huge investment goes into new sub-sea cables, metro area networks and new technologies like photonics that offer the prospect of huge speed and capacity improvements in switching and connectivity within data centres. In some instances both the data centre operators and the mega-scale players invest directly in these developments rather than just buying the output later. The irony is that, where London, Amsterdam and Frankfurt were the places to be if you needed connectivity, these new developments are making that less critical for all but the most latency hungry global applications.
Building out these capabilities is very capital intensive. Even someone with as deep pockets as Google has pointed to the hit in up-front profits that come from such investments. The continued uptake of Cloud computing, in all its forms, underpins the critical importance of this infrastructure to businesses all over the world. This won't be an environment where small local players will thrive in the long term. For sure there is an extremely long tail of these smaller players, catering to local market requirements that are not about to die in the next few years, Data centres aren't a utility yet, but just as electricity generation moved from a cottage industry to one dominated by a small number of providers, data centres and the computing equipment that goes in them are surely heading in the same direction. How that affects consumers is something to look at next.
This post first appeared on the old Cassini Reviews website.