Database abstraction layers, such as the ones we have been discussing, are true APIs in their own right. They are perhaps best thought of as data APIs, because they make applications feel like databases. But data APIs provide the same programming interface capability as REST-based APIs would. What is different about data APIs is their use of the common framework of tables, rows, columns, and SQL queries. Contrast this with conventional APIs’ use of arbitrary frameworks using specific entities and attributes, and an array of callable functions to read or write to the underlying data.

Interestingly, the “primitives” of data APIs (i.e. SQL SELECT, UPDATE, INSERT, and DELETE) match the primitives of REST APIs, which are the basic GET, PUT, POST and DELETE verbs of the web’s HTTP (hypertext transfer protocol) standard, respectively. We mention this not for purposes of technology trivia, but because it underscores the fact that the primitives of today’s REST APIs are directly map-able to the data API metaphor.

The difference is that the data API metaphor has a huge ecosystem of developers, skillsets, BI tools, and SaaS apps already trained in, and compatible with it. Such a pre-existing ecosystem is the non-trivial justification for data APIs. Given the widespread existence of skillsets and compatible tools, using data APIs for data integration is the only real common-sense solution.

Not only are data APIs already data-driven and thus well-suited to facilitate integration for the purpose of data analysis, but they provide efficiencies that no other approach can touch. Even those who feel tabular data is not the best or most intuitive metaphor will admit that as a result of the broad adoption of SQL and database technologies in general, data APIs enjoy a sort of “incumbency” that cannot be matched by others.

Adoption of the data API approach creates a data-on-demand infrastructure around numerous applications. The curated collection of connectors creates a single source of truth around an otherwise eclectic collection of applications, and their otherwise disjoint data sets and databases. The data API approach takes the fragmented and dispersed collections of data that are the result of cloud innovation and corrals them into a coordinated collection of data sets that can be queried and joined in a unified fashion. From entropy comes alignment, manageability, and a sense of order, without losing functionality or precision.