Data Modeling i API First vs Canonical Data Modeling

[#1]

Hi,

In my current program we are planning to undergo a IT transformation. So far the systems have been point to point interfaces and now we want to have a standard API first approach where all systems publish/ consume APIS exposed on API gateway.

As part of this strategy we are supposed to identify all the interface and define API contracts.

I would like to understand that whether we are supposed to find integration needs and come up with a canonical data model or there is a different approach for APIs?

All inputs are appreciated.

[#2]

"Canonical Data Modeling" and "API-first" approaches seat in opposite corners and drive different mindset.

API-first starts with the end in mind. Focus on the consumers and consumer needs rather than on the data and services available in the backend system.

Any Canonical Data Modeling approach will do the opposite and will drive a existing-data/existing-services based mindset that will carry strong influence over the resulting API design.

If you are really looking to design great APIs, set the right mindset, and go with API-first first approach.

[#3]

I'd add to what my colleague Maudrit is suggesting, API-first approach applies similar patterns from DDD (Domain-Driven Design). In which instead of trying to build a single monolithic API, meant to abstract the entire organization and all business units within it, breaks it down into smaller and more independent manageable domains or bounded contexts that are easier to conceive, implement, manage, and improve. So, the single canonical model becomes Multiple Canonical Models, if you wanna call it.

(Commercial) - Apigee has Blueprint Professional Service practice, meant for guiding our customers to navigate everything related to designing and building of great APIs.