Plumbing matters

For anyone who believes that the open Internet and cloud computing are simply dependent on the exchange of ideas and content services, June 30^th provided a timely reminder.  On that day, as blogger TechnoLlama reports, a significant part of the ‘cloud’ failed when a key Amazon data centre (EC2) in Virginia was hit by a violent storm, knocking out its power.  This particular cloud centre hosted important content for various Internet services, including Netflix and Instagram, putting them offline for up to six hours.  Nor was this a wholly isolated incident: the same data centre had crashed earlier in the month, and it seems that cloud outages are now becoming so frequent that websites and Twitter accounts have sprung up to document them.

A central assumption behind the growth of cloud services is that storage and processing are distributed amongst different data centres, in theory making outages less likely. Their whole principle of distribution rests on the assumption of network resilience: if one server is knocked down, others can take up the load. Indeed, this is the one of the founding pillars of the Internet as we know it.  However, the reality is that, from a distributed model, we have been migrating content to more and more centralized services – in terms of both geography and industry concentration.  It’s reported that the top 10 cloud providers are now all based in the US, and that Amazon alone holds an estimated 15% of the cloud market

 
For TechnoLlama, the danger is that this growing reliance on fewer providers has made legal or regulatory control of the Internet an easier task.  For our purposes, the simpler, but no less vital lesson is that the ‘information superhighway’ relies crucially on the resilience of its road (network) infrastructure. Put another way, plumbing matters!