[DZone Research] API Management: Integration, APIs, and OSS Tools

This article is part of the Key Research Findings from the DZone Guide to API Management: Comparative Views of Real-World Design.

Introduction

For this year's API Management Guide Survey, we received 1,287 responses with a 55% completion rate. Based on these responses, we've compiled a quick article on why respondents use APIs, the integration approaches they use, and the prevalence of open-source tools. 

Integration Practices and APIs

In our previous article, we took a look at some of the hurdles that developers face with APIs and why API integration simplicity is so important, so now let’s take a look at how developers are using APIs. When it comes to the types of systems that get integrated, 37% of respondents integrate mobile systems, 37% integrate/analytics systems, and 35% integrate CRM systems. As these were the three most popular responses among this year’s survey respondents, let's use these systems as a means for comparison.

When we look at why developers and/or organizations implement APIs, allowing customers to integrate with their software came in as the top concern for developers and organizations integrating for mobile (57%), BI/analytics (59%), and CRMs (55%). This is where the unanimity ends, however. The second highest concern among respondents working with mobile integration was the need for mobile support (49%), whereas BI/analytics (42%) and CRM (41%) developers told us encouraging users to develop new solutions using their software was the second largest reason for implementing APIs.

Comparing these same three systems to the approaches to application integration that survey respondents use, we again see some interesting trends. The top approach to application integration proved the same among these three categories. 77% of respondents working with mobile system integration, 75% of respondents working with CRM integration, and 74%of respondents working with BI/analytics integration told us that sending messages among applications was their top approach to application integration. And, again, respondents’ choices diverged after the top choice. 61% of mobile respondents reported sharing a common database among different applications as their approach to application integration, whereas 59% of BI/analytics developers and 59% of CRM developers reported that transferring files among different applications was their preferred approach.

Open-Source Tools

Now that we’ve covered the systems and APIs respondents use, let’s take a quick look at the popularity of open-source frameworks and tools. While the popularity of open-source tools is not in itself surprising, it is intriguing that for every tool we asked survey takers about (save iPaaS solutions), open-source solutions took the cake. The two most popular integration frameworks, suites, or ESBs reported by respondents were Spring Integration (40%) and Apache Camel (24%). For messaging tools, the top three choices reported by respondents were, again, all open source. Apache Kafka (39%) proved the most popular, followed closely by RabbitMQ (36%) and ActiveMQ (28%). And Swagger was the most popular framework among respondents for designing and documenting RESTful APIs (45%) as well as API management (68%).

Conclusion

Given the popularity of these open-source tools among our survey respondents, let's take a quick look at how these tools perform on GitHub.

• As writing this article, Apache Camel had 2.1 thousand stars on GitHub and, while Spring Integration iteself doen't have a GitHub repo, the Spring Integration Samples repo has 1,400 stars.

• Apache Kafka has brought in 9.5 thousand stars, RabbitMQ has close to 3 thousand stars, and ActiveMQ has1.2 thousand stars. It seems Kafka has a better showing in the open-source community writ large than in our API Management survey. 

• And, the king of them all, Swagger, has a whopping 12.7 thousand stars. 

By and large, our survey responses seem to mirror the wider developer community's feelings on these various tools for various integration projects. While there were some fluctuations (see the Kafka/RabbitMQ/ActiveMQ comparison), and this isn't the most full-proof method, it seems a safe deduction that the findings presented in this article reflect larger trends in the developer community.  

This article is part of the Key Research Findings from the DZone Guide to API Management: Comparative Views of Real-World Design.