What is the Software Development Lifecycle (SDLC)?
The cycle of life is a useful metaphor to apply to technologies. Like human beings and animals, hardware and software enter the world, do their thing, and then pass on, but not before a new generation arrives to take their place. With applications, the cycle of life is known as the Software Development Life Cycle (SDLC). The SDLC is a project management framework comprising multiple stages: from investigation of problems and planning through coding, deployment, updating, and retirement. This article explores how the SDLC works, its component parts, and the methodologies used in its execution.
It’s worth taking a moment to distinguish the SDLC from the software product lifecycle, which is a related but different process. A software product is introduced to the market, and then hopefully enjoys a “life” that includes being updated with successive versions—before reaching “end of life,” when the product is taken off the market. In contrast, the SDLC refers to the software development and deployment process.
How the SDLC works
The SDLC is about creating software and guiding it through its life of deployment and retirement. At a higher level, though, the SDLC is about solving business and technology problems with software applications. And the process will ideally solve those problems in the most efficient and high-quality way possible.
The SDLC process brings together stakeholders from across an organization. These include software developers, of course, but also business analysts, project managers, line of business managers, testing and quality assurance (QA) teams, and IT operations. Security teams usually have some influence over the SDLC, as do network and storage managers.
Stages of the SDLC
The SDLC has seven stages, though some say there are eight. (More on this below.) Or, there may be fewer, depending on the organization. For instance, some teams may combine planning and requirements.
Identifying business and technology problems is where the SDLC begins. A financial services firm might want more intuitive stock trading software. A distribution company might need to make its warehouse more efficient through software. Realizing that there is a problem to solve is the spark that sets off the seven phases of the SDLC.
Planning—Stakeholders come together to discuss the project’s scope, write project plans, schedule budgets, and so forth. Procurement may get involved if the project requires new development tools or software licenses.
Requirements gathering—Business stakeholders get into a dialogue with business analysts and technology professionals to determine the requirements for the software application. If the project is for updating an existing system, the focus of the conversation is on that system’s deficiencies and how to correct them.
Design and prototyping—At this stage, software developers transpose the requirements into a software design plan. The plan specifies the application architecture and identifies the technologies that will be used, e.g., Java on Linux with MySQL. This stage should also include API planning beyond the technical design. For example, it’s important to understand why the API is being built and how it will impact the business.
Development—This stage is where the actual coding occurs. In modern software, this process may also involve connecting code with application programming interfaces (APIs) that offer up data and functionality from other systems. Development may also mean embedding pre-written open-source software (OSS) code components. When this phase ends, there should be a functional piece of software ready for testing. In agile or DevOps scenarios, that piece might be an incremental addition to an existing application.
Testing—A critical step in the SDLC, testing is meant to identify functional or security problems in the code. The goal is to ensure quality. Test methods may include unit testing, API testing, integration testing, security testing, and performance testing. As issues or bugs are discovered, they are sent back to the developers for remediation.
Deployment—When the software is ready, having been tested and debugged, it is deployed onto production servers. This process often involves packaging it and configuring server software, databases, and the like. Often, it’s an automated process, especially in DevOps and continuous integration/continuous delivery (CI/CD) workflows.
Operations and maintenance—Once the software is deployed in production, it needs to be monitored for performance problems and security vulnerabilities. Sometimes, if demand gets too high, the server running the software will get “hot” and the software’s response time will slow down. IT operations teams should notice this and rectify the problem by spinning up a additonal instances. Additionally, as security vulnerabilities are revealed in the code, or in supporting software like server operating systems and databases, the IT ops team needs to patch the software. This should be a continuous process, with constant monitoring and fast responses to any problems that arise.
The unofficial eighth step of the SDLC is retirement. No software lives forever, so to speak. As it ages, stakeholders will find problems that need to be corrected, usually in a new version of the code. Or, they will decide to create an entirely new replacement application. The IT ops team needs to uninstall the old version and deploy the new one. That’s the true lifecycle. It never ends.
The SDLC can adapt to several predominant SDLC models, each of which aligns with a different software development methodology. The “waterfall” methodology is the oldest and simplest. In this approach, the stages of the SDLC flow in sequence, with one starting after the last one has finished. Planning begins and ends, kicking off requirements gathering, which starts and finishes, triggering development, and so forth.
In contrast, an SDLC based on the agile software development methodology comprises a series of ongoing release cycles. Each cycle produces a small, incremental change over the earlier release. It’s iterative and cyclical, with the SDLC repeating and going back on itself in loops. This approach enables stakeholders to identify issues in a project and resolve them before they mature into more serious problems.
The “lean” software development methodology creates an SDLC that mimics lean manufacturing principles, which encourage an efficient flow in work processes and continuous improvement. In an “iterative” SDLC, each cycle of development results in the deployment of an incomplete version of the app. The first iteration delivers a small subset of the requirements. Each iteration after that adds more requirements. By the final iteration, the app will embody all its requirements.
A “spiral” SDLC offers a development process driven by a project’s distinctive risk patterns. Stakeholders evaluate the project and determine which elements of the other process models need to be incorporated. A “V-shaped” SDLC involves running verification phases and validation phases in parallel. Each verification phase links to a validation phase, resulting in a v-shaped plot to describe the workflow.
The SDLC never stops evolving. As new modes of software development emerge, they will have an impact on the SDLC. The underlying goals will never change, however. The SDLC is always about producing the best quality software in a reliable, fast, and efficient process.