Use agile building methods

In 1960–1995, monolithic information systems were mainly built, in which data were mostly shared within the organisation and in some cases with subcontractors. Digital services were typically used on a desktop.

As the Internet evolved, a new, decentralised way of processing data came about. In the late 1990s, Service-Oriented Architecture (SOA) became popular. It aimed for efficient re-use of programs and data in business functions.

Key points of service-oriented architecture include:

• service orientation
• interoperability
• loose coupling.

These aspects make modifying information system components faster and easier.

Service-oriented architecture is based on software that offers a range of services with four key features:

1. the service carries out business functions with a specified goal
2. the service operates independently
3. the service hides its internal complexity
4. the service may consist of other services.

In service-oriented architecture, services communicate through APIs.

Microservices, which have become more widespread since 2010, also operate through APIs but are more independent and lightweight solutions than SOA services.

Microservice architecture and the API First model have enabled faster and more flexible development. Smartphones have simultaneously emerged as a new user interface, significantly increasing user numbers.

Microservices and the API First model make systems more flexible and easier to update. Thanks to them, new features can be implemented faster and more cost-effectively than in large monolithic information systems.

Serverless architecture, which has become more common since 2014, allows automated management of server resources by a cloud service provider, such as AWS. This speeds up development and reduces costs as resources are only paid for per use.

The use of serverless models has increased dramatically in the 2020s. In these models, code is executed as needed (such as FaaS), and many serverless solutions use API calls or are transaction based.

To sum up, we can say that the importance of APIs has increased continuously in recent decades.

In API design it is important to account for granularity, which means the extent to which an API divides functionality and information among individual resources.

• An excessively coarse-grained API may return too much data (over-fetching), which slows down its use and makes it ineffective.
• Similarly, an excessively fine-grained API may result in a situation where a customer has to make several separate calls to obtain the information they need (under-fetching), which adds to complexity and latency.

Balance can be found through customer-oriented design:

1. by listening to user needs
2. by working together with partners
3. by providing clear documentation
4. by managing versioning.

With correctly chosen granularity, the API is easy to use and efficient, and it can be expanded to meet future needs.

A well-designed API improves the efficiency and long-term sustainability of software development. The API First model also supports software testing and quality assurance from the very start of a procurement.