Piecemeal adoption of cloud services and apps has created many collections of stovepiped services that don't talk to each other. More and more users — especially business users — have started grappling with how to connect these cloud islands together. Answering this requirement is a necessary part of the maturing of the cloud model. The answers that are starting to emerge suggest that there may be some innovative approaches in the works that will make us think afresh about how and where the connections should be forged.
Joining the dots in the cloud ought to be easier than traditional application integration because all good cloud services are inherently designed to connect, and those connection hooks (technically known as APIs) have to be highly standardized and web-friendly to be any use in a cloud environment. But until recently, taking advantage of that ease of connection has had to be a hand-crafted effort. Tools to automate the process have been relatively rare. Sure, there are plenty of data integration options, but the Web lends itself to other, more loosely-coupled forms of connection. Those tools have been few in number, but in the past few months, there's been a sudden explosion of new options.
Last week's announcement of Amazon Simple Workflow Service (SWF) was the most recent, making it easier to orchestrate different components of an application that may straddle cloud, mobile and on-premise IT assets. This is a clever commercial move by Amazon, because this capability further encourages the type of loosely coupled application architecture that makes best use of Amazon's cloud infrastructure, as well as making it easier to use Amazon-hosted assets alongside existing on-premises IT. But it's still a relatively low-level application messaging tool that targets developers rather than typical users.
The great thing about the Web is that it allows you to break away from the old three-tier model of application integration that insists on a heavyweight layer of middleware to bind everything together before it can be acted on. In some ways, SWF doesn't change that model, it just puts the heavyweight parts behind a cloud API that developers can call on instead of having to build it themselves. Flotype, a startup backed by Salesforce.com and Andreessen Horowitz, is in a similar vein.
But connection doesn't always have to be engineered down in the bowels of the infrastructure. There are plenty of other examples that make it easy to link services with a lightweight mashup or by bringing filtered information together on demand in the browser for rapid analysis and interaction by the user.
One of the hot new start-ups drawing attention around the turn of the year was IFTTT, which stands for IF This Then That. Users can set up conditional events that trigger simple actions in popular activity streams and apps including Twitter, Facebook, DropBox, LinkedIn, RSS, email, SMS and even voice calls. It's mainly used to automate the replication of routine tasks between online services, such as posting to multiple streams, archiving items posted online or setting up alerts.
An example that uses the cloud to provide user-centric workflow, running across transactional applications and document collaboration, is RunMyProcess [a client for whom I recently wrote a white paper: see disclosure]. Here the emphasis is not so much on connecting applications as on moving from applications to documents and on to people and back — what in the white paper I called the last mile of business automation. Often the aim is simply to finish automating the gaps in an existing process that straddles several different application stacks rather than creating a completely new process. A cloud platform finds this easier than more traditional approaches because it is better able to connect to multiple touch points, whether in the cloud, on in-house servers, or on individual client machines.
Cloud services and apps make many activities less costly and cumbersome to automate, but unless those newly automated activities can be connected up into automated processes, many of the potential productivity gains and cost savings will be squandered. The good news is that there are many different types of solution now emerging to help make those connections. The less good news is that it will take some learning and experimentation to figure out how best to use those tools — and the bad news is that too many people will try to connect cloud services using the same techniques that worked for them in the on-premise world, whereas it will take new thinking and new approaches to realize the full potential of the connected cloud.