The API is the Center of the Application (Integration) Universe

#mobile #fasterapp #ccevent Today, at least. Tomorrow, who knows?

Some have tried to distinguish between “mobile cloud” and “cloud” by claiming the former is the use of the web browser on a mobile device to access services while the latter uses device-native applications. Like all things cloud, the marketing fluff is purposefully obfuscating and sweeping under the rug the technology required to make things work for consumers, whether those consumers be your kids or IT professionals. Infrastructure is not eliminated when organizations take to the cloud nor do the constraints of web-based protocols and methodologies become irrelevant when Bob uses a service to store photos of his kid’s piano recital on Flickr.

The applications and web browsers on a mobile device are using the same technology, the same protocols, suffering under the same constraints as the rest of us in wireline land. If developers are as smart as they are lazy (and I say that as a compliment because it is the laziness of developers that more often than not leads to innovation) they have already moved to an API-centric model in which web site and device native-app interfaces both leverage the same APIs.

This isn’t just a social integration phenomenon – it isn’t just about Twitter and Facebook and Google. API usage and demand is growing, and it is not expected to stop any time soon. Given the option, developers asked about desire to connect to services (assuming service = API) the overwhelming response was developers would like to connect to “everything, if it were easy.”  (API Integration Pain Survey Results)  

The API is rapidly becoming (if it isn’t already) the center of the application (integration) universe. This unfortunately has the potential to cause confusion and chaos in the data center. When a single API is consumed by multiple clients – mobile, remote, applications, partners, etc.. – solutions unique to each quickly seem to make their way into the code to deal with “exceptions” and “peculiarities” inherent to the client platform.

That’s inefficient and, when one considers the growing number of platforms and form-factors associated with mobile communications alone, it is not scalable from a people and process perspective.

But reality is that these exceptions and peculiarities – often times caused by a lack of feature parity across form-factors and platforms – must be addressed somewhere, and that somewhere is unfortunately almost unilaterally determined to be the application. Do we need to treat mobile devices differently? In terms of performance and delivery concerns, yes. But that’s where we leverage the application delivery tier to differentiate by device to ensure delivery. That’s the beauty of an abstracted, service-enabled data center – there’s an intelligent and agile layer of application delivery services that mediates between clients (regardless of their form factor) and services to ensure that delivery needs (security, performance, and availability) are met in part by addressing the unique characteristics and reality of access via mobile devices.


This is exactly the type of problem application delivery is designed to address. Multiple clients, multiple networks, all accessing the same application service or API but requiring specific authentication, security, and delivery characteristics to ensure that operational risk is mitigated in the most efficient manner possible.

This includes the ability to throttle services based on user and client, a common approach used by mega-sites such as Twitter. This includes the ability to provide single sign-on capabilities to all clients, regardless of platform, form-factor and support for enterprise-grade authentication integration to the same API or application service. This includes leveraging the appropriate security policies to ensure inbound and outbound security of data regardless of client, such that corporate data is not infected and spread to other consumers.

A flexible, scalable application delivery tier addresses the problem of a single API being utilized by a variety of clients in a way that precludes the need to codify specific functionality on a per-platform or form-factor basis in the application logic itself, making the API simpler and easier to maintain as well as test and upgrade. It makes APIs and application services more scalable in terms of people and processes, which in turn makes the development and deployment process more efficient and able to focus on new services rather than constantly modifying and updating existing ones.

Service-oriented architecture may have begun in the application demesne as a means to abstract and isolate services such that they could more easily be integrated, maintained, and changed without disruption, but the concept is applicable to the data center as a whole. By leveraging SOA concepts at the data center architecture level, the entire technological landscape of the business can be transformed into one that is ultimately more adaptable, more scalable, and more secure.

I’ll be at CloudConnect 2012 and we’ll discuss the subject of cloud and performance a whole lot more at the show!


Published Jan 23, 2012
Version 1.0

Was this article helpful?

No CommentsBe the first to comment