WSO2 sponsored this post.
The shift in how we live and work, which started with the pandemic, has brought an even stronger push by many organizations to scale operations across cloud native and distributed environments. To survive and thrive, companies must now seriously look at cloud native technologies such as API management and integration solutions, cloud native products, integration platform as a service (iPaaS) offerings and low-code platforms that are easy to use, accelerate time to market and enable re-use and sharing.
In this article, we look at how low-code development platforms across the application development lifecycle can deliver enhanced digital experiences.
Modern applications share many key requirements such as authentication/authorization, logging, notifications, monitoring and deployment options. As these applications become widely used and deployed, ensuring scalability and availability has also become equally important.
In the past, most tech-centric companies implemented these application requirements from scratch, as they usually had the technical expertise and the know-how to do so. The problem is that many of today’s businesses do not have all the skills and resources. Data shows that many digital transformation projects fail due to cloud complexity, an unmanageable infrastructure footprint and a lack of IT system agility.
For context, the U.S. Bureau of Labor Statistics projects employment of software developers to grow 22% from 2019 to 2029, much faster than the average for all other occupations stated. More alarmingly, the bureau expects a shortage of 1.4 million developers in 2021. A McKinsey & Company 2018 report found that although 80% of enterprises have attempted a digital transformation project, fewer than one-third have succeeded at improving company performance. Another report, the 2019 Chaos Report from The Standish Group, found that 84% of technology projects partially or completely fail.
So there is a clear need to get more people involved and simplify application development.
Enter, Low Code
Cloud computing has provided a whole new approach to developing software where some common functionalities are provided as services. These systems are well-tested, scalable and readily available. This has allowed developers to solely focus on business problems and let other services handle all other complexities of the system to make the applications run reliably.
In the end, the required application code could simply be an assembly of already-existing services. A developer can just bring them together to provide the required experience to the user. This is what low-code software development promises.
A low-code platform enhances collaboration between business domain and technical experts who can all help to create applications. Hard-core coding requirements can be handled by traditional developers. Other high-level workflows can be modeled and implemented by domain experts, who can use a graphical form to define the data flows and processes.
So, the need for low-code development is real. However, in the past, developers stayed away from this style of development due to limitations such as focusing mostly on a configuration-based domain-specific language (DSL), limited options to express certain scenarios compared with code-based approaches, no proper CI/CD support, lack of debugging and vendor lock-in.
To overcome these, we propose a low-code software development lifecycle, which will cover all these concerns and provide the same functionality as in traditional code-first practices while providing increased agility in a low-code environment. This is broken down into the following stages of application production.
An ideal low-code software development lifecycle and platform should address these concerns and provide the same functionality as a traditional code-first approach, while providing increased agility in a low-code environment. The following looks at such a proposition.
The above diagram shows how each stage of the low-code development lifecycle is related. The main theme of the flow is iterative development. There are certain stages or a collective of stages that are used iteratively until it is finalized.
Let’s dive into the details.
Design and Architecture
This is the first step in our development lifecycle. Here, we gather all the system requirements and come up with a high-level solutions architecture. Typically, this stage is done by solutions architects, where they will draw L0 and L1 architecture diagrams to provide details on the high-level components. The following lists some of these components:
- API gateways
- Database servers
- Message queues
The main benefit of this step in our environment is that it is directly associated with the low-code development environment where we will be developing our application. It is no longer in a separate document that the developers read later. Rather, this is integrated into the same experience everyone is involved in. This solution architecture is reused in other aspects of the system, such as defining scaling requirements of the system, communication links between components and defining observability requirements. These are directly documented in our low-code platform in a declarative manner, and they actually become functional features in the final deployment.
In the development stage, we basically drill down from the previous design and architecture view to further define its components. In a typical service or an API, this would be defining its functionality or business logic. The user should be provided with a graphical environment to define its execution flows.
To facilitate the implementation scenario and select the type of developer, we should have two views for the logic we are developing: a code and a low-code mode. The low-code mode is where most of the developers will be defining the execution flows. A developer will be presented with a set of connectors, operations and other actions such as data mapping and validation.
The low-code developer will be able to handle most of the business logic using the graphical modeling approach. However, there can be scenarios where certain logic can be more easily expressed in direct code. This is where we need to have a direct code view of the logic we are implementing. This way, we can quickly switch over to a code view and create a custom function to implement the required behavior. By doing this, we get the best of both worlds and none of the drawbacks of either approach due to the seamless mapping between a graphical model and its code representation.
In the development stage, certain activities can be made more intuitive to the user with the use of artificial intelligence (AI). Aspects such as data mapping and flow activity suggestions can be done using AI with the use of historical usage data and other knowledge of interfaces and outputs of actions in the system.
The testing stage goes hand in hand with the development stage. Here, all the execution flows are tested using unit and integration tests. The system should provide integrated features to automatically generate test scenarios from the services and APIs developed in the previous stage. It should allow creating test plans to execute certain scenarios in order and to define dependencies between each execution flow.
The tests will be continuously executed with every change done to the implementation of the flows and defined in the continuous integration rules. The development and test stages together will be going through an iterative process until the code flows are finalized.
In the deploy stage, the user should be able to select a deployment environment where the application should be deployed. This can be a cloud provider such as AWS, Azure or Google Cloud or an on-premises location in your own data center. The hybrid-cloud environment is important since not all businesses would be willing to host all their applications in the cloud due to security and regulatory restrictions.
After a deployment provider is selected, the user should note low-level details such as if it uses Kubernetes underneath and what its configurations would be or if any serverless platforms are used. The details also should provide simple configurations for features such as auto-scaling, high availability and service discovery.
A continuous deployment strategy also should be configurable as well at this stage. It may be the case that after each change in the execution flows, after tests are executed, a new deployment version would be deployed to production. Or we can have another intermediate sandbox or a pre-production environment to test out the changes manually and to promote the environment to production.
The test and the deploy stages are executed iteratively until we optimize the deployment parameters to make sure everything is working optimally.
The observability features of the platform enable developers to monitor the behavior of the application after it’s deployed. In the lifecycle stages we defined here, this can be done in multiple levels, first in the design and architecture level to provide a systemwide view of the components working together. It can provide information such as the availability of services, their uptime, request/response statistics and error rates.
Then, we can drill into individual components to get a finer view of the flows that are implemented. The low-code view provides a natural view to this, where we can overlay the user-defined flows to provide request or transaction-level information such as tracing details, or else aggregated information for each action in a flow, such as its success rate and average execution time.
The ability to drill down from the high-level deployment view to each action in a flow provides us with powerful capabilities to debug, optimize performance and implement predictive maintenance of the system.
Organizations across sectors have had to accelerate their digital transformation efforts, not only to streamline and make operations more efficient, but also to respond to dramatic fluctuations in demand and deliver enhanced digital experiences.
In this post, we discussed the need for every organization to become a digital business, and the challenges we face in the current times for achieving this. A low-code-based development strategy, along with a suitable development lifecycle is critical to meet these demands.
The ideal offering should come with a low-code and code hybrid development environment where the platform’s graphical modeling generates clean code and the changes flow bidirectionally without any additional code-level annotations or additional configurations. It should also come with hybrid-cloud support and no vendor lock-in, where its execution flow code can be independently compiled and run in your own local environment if needed.
If you want more details about an IPaaS solution and want to see one in action, please visit Choreo, a comprehensive platform for low-code, cloud native engineering.
Featured image via Pixabay.