In today’s competitive landscape, businesses need to make decisions quickly; whether it’s a new marketing campaign, a new product enhancement, a new partner portal or an employee productivity app, businesses are competing on speed.
What is needed is flexibility. And what better way to do this than to create purposeful, agile application building blocks that can be easily pieced together with well-designed, well-managed APIs to quickly create what is needed.
Connecting these building blocks results in a network of interchangeable functionality — a network of sub-functions, or applications. We call it the application network. Let’s future-proof ourselves and hedge against uncertainty by creating these building blocks, and then allow for the flexibility to rapidly piece them together on an on-demand basis.
The Application Network
The application network is the future. It emerges from the creation of multiple API-enabled microservices; connecting these microservices with a strategic API approach results in a composable network. The network allows the flexibility to rapidly piece together different services for multiple functionalities an on-demand basis, providing business agility and a robust platform for innovation.
What is application building block
An application network is composed of application building blocks. These have multiple elements. and it’s critical to separate the concerns between each. The API interface, the API implementation, and the API management aspects all have their own specific, unique lifecycles to follow. This building block should itself be treated as a product since these characteristics are common to what a good product should also have.
Therefore, it makes sense to treat a building block from a product-centric approach. We see this product-centric lifecycle as having three distinct stages: design, implementation, and management.
Designing an API
Designing an API starts with an acknowledgment ahead of time that the API production process will start from an outside-in perspective, beginning with the “interface/contract” of the API. That is to say, first, let us decide what the API will look and behave like before we actually begin to code the backend logic.
Often, developers build APIs without knowing the API has been validated and accepted. As a result, there is always this air of uncertainty: “is this what my consuming developers want?”
API developers must design the “user interface” of the API first — this is also known as the API contract. This approach is typically known as a “design-first” approach, and it should follow a deliberate API design lifecycle in order to optimize for the best API experience.
As a result, it is important to be able to do this in a human-readable fashion — to specify the contract in a way that humans can easily digest.
- Identify process and business requirements
- Create logical data model
- Translate into logical service / API groupings
- Model API resources
- Model API operations/methods
- Model request/response payload / codes
- Mock up the API
- Publish interactive console
- Create Notebook use cases
- Receive developer feedback
- Modify API design as appropriate based on developer feedback
- Continue to validate
The process of API validation
At this point in the process, the API designer (i.e. the designer of the API contract) is ready to have the API be validated and tested by the API consumers (i.e. a client mobile app /website developer, or another API provider in some cases).
The currency of conversation between both parties will be via interactive tools such as the API Notebook and interactive documentation. This process of validation may be brief or extended over numerous iterations.
What is API Notebook
Think of API Notebook as the artifact to convey the inspiration for what is possible with an API. It serves as a client application — it calls one or more APIs; it is, therefore, live use case in action, where one can easily mash up multiple APIs, experiment, tinker and see what can be built. Another benefit of having a Notebook is that it also serves as a sort of testing sandbox for an API.
Repeatable design of an API
Any well-designed API will have repeatability in it as well as repeatability across other APIs. This can easily be encapsulated into best practice patterns both at the structural level of the API (nouns/resources), as well as at the method level (verbs). So as API developers go about the design process, it is important to be able to discover and share repeatable patterns.
Systematic approach to API implementation
API implementation is a critical piece of enabling
a next-generation enterprise. Enabling for dozens, potentially even hundreds or thousands of APIs to be connected down to a backend and connected to each other will be key. This must be done in a systematic manner (as opposed to point-to-point code).
The criteria for systematic API implementation
- Data mapping
- Connectivity to popular SaaS, on-premise systems and data, file and other protocols
- Out of the box
- Extendable and based on best practices
Creating API best practices repository
Benefits of a best practices repository:
- Increase business agility
- Share best practices with reusable templates and logic
- Leverage best practice patterns
- Rapidly deploy APIs: fail fast, succeed faster
- Minimize point-to-point logic, and future proof for stability
Best practice patterns to be leveraged when creating new building blocks.
- popular database calls
- most commonly done transformations
- most popular workflows across multiple systems
- connections to popular SaaS and on-premise endpoints
- REST-SOAP transformations
Testing the API implementation
At this point in the process, the API provider (i.e. the developer behind the API implementation) is ready to have the ‘guts’ of the API to be tested. Test automation tools are critical here, as this integrates into the DevOps processes of continuous delivery and deployment.
The importance of DevOps for API management
Embracing modern DevOps-centric processes and tooling is critical to reducing time-to-production, and this should apply to your application building blocks as well. Once the application building block has been assembled and tested, deployment should be as easy as the click of a button.
The use of a hybrid integration platform that is lightweight, easy to install, and suitable for CICD workflows is key. The ability to have seamless support for dependency management, testing, version control, and automated deployment tooling should be an assumption.
API security is everything
It is critical to ensure your application building blocks are following best practices in security and architectural governance by applying policies to them at runtime. Monitoring all traffic is equally critical because it just takes just one weak link to bring the ship down.
Policy configuration examples
- Traffic management (e.g. rate limit)
- Access policies (e.g. OAuth2)
- Identity policies
- Custom policies
Monitoring and Analytics examples
- Infrastructure logs
- Service uptime analysis
- Client consumption data
- Provider analytics
The discoverability of your API
Imagine your company with hundreds — if not thousands — of APIs in your expansive application network. Imagine you’re adding several new ones every day. Being able to appropriately publish them so the consuming developer can find, research, and understand them easily could make or break your entire program.
There is no point in building something that won’t ever be found, let alone used.
Building blocks will change. It’s a WHEN, not an IF. So be ready for it, which requires a carefully planned set of policies, procedures and the right platform to seamlessly migrate clients across new versions of APIs. Getting this migration wrong will affect your customers. That’s not a risk worth taking.
It’s key to have the ability to adapt towards the new operating model, one where DevOps and lean practices are adopted, as well as the creation of new roles and responsibilities to support this new operating model.
Who’s going to make the new API
Depending on the maturity of your organization, there may be one person handling all these responsibilities or there may be multiple people doing so. The important thing to note is that each stage in the lifecycle provides specific value, which ensures application building blocks provide desired business outcomes. There can be different people working on a new API on each level — API Analyst, API Admin, Application Network Architect, Integration Developer, API Manager, etc.
CloudGeometry can help
Whether you need help to integrate a new cloud service into your existing business infrastructure or a young SaaS, PaaS or IaaS company who has to add your solution to the clients’ technology stack, CloudGeometry has expertise and resources to help you. We have many years of experience in working with APIs of popular data providers as Facebook, Google, Amazon, Twitter and many big AdNetworks. Those APIs are frequently updated and we have to design and build special flexible systems to quickly change the integration rules without an impact to the downstream systems. We have also been working for many years with legacy industrial system helping our clients like GE Digital, KryptonCloud and Imantics to connect their new PaaS and IaaS solutions. We provide a range of advisory, development and support services helping you to integrate cloud and on-premises system into one stable solution.