28 days ago

Strategy and Requirements for the API ecosystem

Link: http://feedproxy.google.com/~r/Soapbox/~3/db2zuEQc2TU/strategy-and-requirements-for-api.html

Is there a framework or methodology for establishing the business / ecosystem requirements to drive API strategy and development?

At an industry event I attended recently, hosted by a company that sells tools and technologies for the API ecosystem, some of the speakers advised that when presenting to non-technical stakeholders, you need to talk about service value/benefit rather than APIs. But this raises an important question, how to identify and quantify service benefit, and how to negotiate share of value between different players in the ecosystem?

One of the ideas of the API economy is that you don’t have to maintain all the capabilities yourself, but you find other enterprises that can provide complementary capabilities. So you need to identify and understand what capabilities are available, and map combinations of these capabilities against the demands and unfulfilled needs of potential customers. Then having identified in broad terms what capabilities you wish to combine with your own, and worked out where the service boundaries should be, you may select organizations to partner with and agree business and commercial terms, or create a platform to which many third parties can connect. The technical design of the API should then reflect the service boundaries and commercial arrangements.

In the early days of service-oriented software engineering, people always wanted us to tell them how large their services should be. Not just macro versus micro, but broad (generic) versus narrow (specific). To what extent should a service be completely purpose-agnostic – in other words, with no restrictions on how or where it may be used – or does this conflict with other design goals such as reliability or data protection?

The answer is that it depends not only on what you are trying to do, but how you want to manage and govern your service architecture. A broadly scoped, purpose-agnostic service (or service platform) can achieve wide usage and economies of scale, but may be more complex to configure, test and use, whereas a more narrowly scoped context-specific service might be easier to use but with lower reuse potential. Among other things, this affects how much of the service composition and orchestration can be done by the service provider (supply side), and how much is left to the service consumer (demand-side). And even on the supply side, it affects how much work needs to be done by the integration experts (“town planners”), and how much can be left to citizen integration (“pioneers” and “settlers”).

One version of this challenge can be found in large global organizations, working out exactly what functionality should be provided centrally as shared services, and what functionality should be left to local operations. Ideally, the service architecture should be aligned with the business and organizational architecture.

The word “economy” also implies attention to accounting issues – sharing costs and benefits between different players. Although we may regard cloud as almost infinitely extensible, this doesn’t come without cost: if the number of service calls goes through the roof, someone has to pay the cloud provider’s bill. This is already an issue within large organizations, where we commonly find arguments about whose budget will pay for something. And I have seen some great ideas come to nothing, because the benefits were spread too thinly and nobody was able to fund them.

So although vague appeals to innovation and imagination might be good enough for a marketing pitch, serious strategic thinking is about discovering where there is untapped value in your business and its environment, and working out exactly how an API strategy is going to help you unlock this value.


At the CBDI Forum, we were talking about these issues many years ago: our Service Architecture and Engineering® methodology is still available from Everware-CBDI. Here are some of the articles I wrote for the CBDI Journal.

More at  http://everware-cbdi.com/cbdi-journal-archive