Remember the old days of cloud computing, when pundits were predicting that public cloud services would displace every server in every data center? Well, that was a dumb prediction from the first, but it's obvious from the trends in provider revenue that the cloud is exploding. How do we harmonize the truth of cloud growth with the notion that having the cloud displace everything else was a myth?
OTT & the Enterprise
Easy, as it happens, but profound. First, most public cloud services aren't sold to enterprises at all, but to over-the-top (OTT) companies that offer social networking, content, and consumer services. Most of these companies are startups themselves, and the way they've used the cloud is a big factor in how public cloud providers have evolved the Web services they offer. Serverless computing, for example, grew out of how Twitter handles message processing.
Amazon is the principle provider of OTT hosting in the cloud, a likely reason it's the largest provider of public cloud services. Amazon has strong pricing and features, and because it's an OTT itself, it's the biggest consumer of its own cloud and its OTT experiences guide it in optimizing service decisions and features. Not to mention that Amazon's use of its own resources gives it an advantage over many other providers with respect to economy of scale.
Google is the laggard among the top three public cloud providers, and perhaps one reason is that Google would be a direct competitor to some of the OTTs looking for public cloud services. Google also seems to be struggling to define a strategy for the other segment of the public cloud market, the enterprise. At last week's Google Cloud Next, Google had a lot to say about enterprise use of the cloud, but most of it repeated past positioning. It still hasn't found a launch point for a truly new positioning or cloud model.
For the enterprise, the cloud isn't displacing stuff, it's augmenting it. The majority of enterprise use of the cloud, so enterprises themselves tell me, is building more agile front-ends to legacy applications. Think Web- or mobile-enabled apps. This approach works because most businesses have specific transactions they process, representing the buying and selling and employee management tasks that make up their business processes. These change less than the way workers, customers, and suppliers use or supply information, and the cloud is a good way to speed-match between the two processes.
Nearly all enterprises think their use of public cloud services as a front-end to their current applications is likely to grow, and grow significantly. Microsoft has been the leader in this aspect of public cloud computing, owing in large part to the fact that it can position its Azure cloud services and Windows-based data center and desktop tools as a single application continuum, something that facilitates the very front-end mission approach that enterprises are taking.
Amazon has recognized the reason for Microsoft's leadership, and has been working to facilitate the integration of its cloud services with the enterprise. Its Greengrass tool lets AWS applications spill out of the cloud and onto the user's premises, making the edge of the cloud a part of an enterprise's own facilities. Amazon also recently introduced Snowball, a tool designed to move large databases between premises and cloud. That's helpful if the cloud front-end needs access to a database, because it eliminates traffic charges and delays if information crosses the boundary between cloud and data center.
Hey, Over Here
The application battleground that ends up uniting the enterprise and OTT spaces is event processing. Events are little units of stimulation that might be generated by a sensor, a management system, or even a worker. Events typically signal that something needs attention, which means that some process has to be started up to handle them. You can't easily predict how often events are generated, so they can't be easily handled by traditional reserved-resource cloud processes. That's what gave rise to "serverless" cloud computing.
Remember that events "signal that something needs attention." It follows that there are some "somethings" that might need attention right now, meaning that they can't accept a transit or processing delay. That makes event processing a challenge for cloud services, because most event sources aren't particularly close to a public cloud data center. Could Microsoft, which has a stake in the data center, use this to gain further enterprise cloud business at Amazon's expense? It could well be true.
Events might be the next application front-end focus for the enterprise. The Internet of Thing's business value lies in large part in the notion that events that were usually recognized and handled by a worker would instead be handled by a cloud process. Obviously that's the next step in productivity enhancement; first you make the work easier for the worker, then you take the work right to the IT process and bypass the worker. Microsoft has a lot more architected cloud services directed at workers and worker roles than either Amazon or Google. Could it create a "worker emulator" role for IoT?
It's easy, seeing the obvious and direct competition between Microsoft and Amazon for the enterprise cloud, to forget Google. Google has the least specific enterprise tools and strategies of any of the cloud providers, but arguably the most edge data center points of any of them. Remember that Google caches its content everywhere. Google also has leading-edge cloud and cloud networking technology. If Google gets its enterprise act together, it might end up being the player that sets the standard for enterprise cloud services and drives the next stage of public cloud adoption.
Follow Tom Nolle on Google+!
Tom Nolle on Google+