As we enter a new decade, APIs are involved in many current technology trends. Web APIs are common in microservices, service mesh, public cloud platforms, machine learning, and smart devices. In this inaugural episode of the APIs Unplugged podcast, Mike Amundsen (MuleSoft API Strategy Advisor) and Matt McLarty (MuleSoft Global API Strategy Leader) discuss the most significant technical trends related to APIs for 2020.
Tech trend #1: API specifications
The first part of the discussion focuses on emerging API specifications. Matt recounts his experience attending the first-ever API Specifications Conference in Vancouver last fall. There he observed a convergence of different API communities: Open API Specification adherents, GraphQL pioneers, cloud nativists with an affinity for gRPC, AsyncAPI evangelists, as well as semantic modelers. The event allowed members of these communities to tout the strengths of their particular protocols, while also showing how much they had in common. Mike talks about the contention between modeling approaches that facilitate long-lasting design — such as ALPs that he co-authored with Leonard Richardson — and more explicit specifications that ease initial implementation but are more prone to obsolescence.
Tech trend #2: Organizational roles and tools
The second part of the discussion explores the relationship between tools and roles in an organization that is adopting APIs for provision or consumption. Mike refers to the book he co-authored in 2018, Continuous API Management, which includes a section detailing the roles that emerge in a mature API organization. His co-author, Mehdi Medjaoui, often talks about there being many roles in an API organization, although the number of people required to fulfill those roles grows over time. Mike also relates his own experience working with customers: “Four or five years ago it was almost all tech, all IT, but now I’m seeing a wider audience in the room. That probably means different tools for different people are required.” Matt closes this section noting the inextricable link between business, technology, and organizational concerns that are evident in the API space.
Tech trend #3: Service mesh
The third section looks into the popularity of the service mesh concept and its relationship with APIs in an organization. Mike points out that “service mesh” has already become an overloaded term, ranging in meaning from the initial “data plane/control plane” definition to any large-scale microservices environment featuring an entangled set of services (a “service mess”). Matt discusses the “smart endpoints, dumb pipes” microservices principle he’s seen be over-emphasized by many organizations implementing service mesh. Mike points out this also relates to those organizations scaling up, noting, “when you get whole parts of your organization, this becomes a network management and a traffic management problem.” They then segue into a discussion on serverless computing, and the potential issues that can arise as its adoption matures in the enterprise. Mike highlights the flexibility of serverless, and how that comes with non-deterministic behavior. Matt builds on this point, adding that, “much of software engineering history has come from a perspective of ‘we can be deterministic about the systems we’re building’ and I think we’ve reached a stage of distribution where everything is non-deterministic.”
Tech trend #4: RPA
The topic of determinism leads to a dialogue on one of the hottest current trends in enterprise computing: robotic process automation (RPA). Matt believes that RPA can be a cop out for enterprises that don’t want to do the hard work of refactoring their legacy environments. Mike remarks that RPA can be seen as glorified screen scraping, but counters that by saying he is intrigued in RPA’s potential to impact back-end APIs. He sees designers eventually needing to cater to machine requirements as well as human developers. Matt emphasizes that API design is important for both types of users. The podcast ends with Mike reiterating his optimism for the impact of RPA on designing service outputs.
Hope you enjoy this episode! Subscribe to our newsletter above for more podcast episode recaps.