Why Things Need Clouds

In the fast-changing computing industry, it seems like every few years there is a new "next big thing" that everyone is excited about.  Sometimes these next big things have proven to be fads or simply repackaged versions of what has come before, but more often than not the new next big things are hot because they bring new capabilities and potential applications.

It now appears that the Internet of Things may be about to replace Cloud Computing as the latest next big thing, but that hardly implies that the cloud is going away.  Some people initially thought the cloud was just a fad, but the widespread and ongoing move to cloud-based applications demonstrates otherwise.  In fact, the value of cloud computing is now too well established for it to remain the next big thing.  The cloud has become vital infrastructure for a host of applications, as it will be for the next next big thing, the Internet of Things.

Of course, cloud computing succeeded for reasons that had nothing to do with the Internet of things.  It has provided solid ROI in cost of services, and has also made services tend to be more reliable because they're run by specialists.  In addition, by consolidating data enterprise-wide, it has enabled new kinds of business intelligence "big data" analytics.  The cloud is here to stay.

That's good, if you're an enthusiast of the Internet of Things, because the IoT needs the cloud to reach its full potential.  By themselves, the "things" on the Internet could be an administrative and maintenance nightmare.  With millions or billions of such objects on the Internet, many of them very small and low-end, it's largely impractical to design systems around a point-to-point architecture.  If a "thing" needs to interact with a half dozen different services, and a service needs to interact with thousands or millions of "things," it's impractical to imagine them all contacting each other directly, and particularly daunting to imagine how the "things" will be administered and updated.  If we want to allow each "thing" to have a single "home base" server, and if we want to minimize the permanent state and data maintained on each device, the cloud is the obvious point of convergence.  

The natural way to connect "things" and services is via the cloud.  Each "thing" can have a single cloud-based service to which it delivers its data, and from which it takes its configuration information.  Then, an arbitrary number of services can interact with that same cloud service to access the data provided by the "thing."  By minimizing the complexity of the "things," we not only make the "things" easier to administer, we also make it easier to build more services, of increasing complexity, that make use of the information from the "things" via the cloud.  The part of a "thing" that interacts with the cloud can be made extremely generic, and therefore relatively easy to manage.

If we're really moving towards a world with billions of intelligent, Internet-connected objects, it's absolutely vital that we make those objects as self-sufficient as possible.  This means keeping configuraiton to a minimum, and moving complexity to the cloud.  This has been one of the secondary benefits of the cloud for office technology, but will be a primary one for the Internet of Things.  "Things" don't need to know much more about the Internet than how to upload data and download data and configuration information.  It's hard to see a rationale for doing that with more than one service.

The bottom line is the services that make use of the Internet of Things are likely to be almost entirely cloud-based.  It will be hard for them to make use of applications and data not already in the cloud.  If you're hoping to make use of the Internet of Things, step number one is to make sure that all your important data and services are available from the cloud, lest you find yourself two next big things behind.