Our highest priority is to satisfy the customer through early and continuous delivery of valuable software.

    —Agile Manifesto

    Continuous Delivery Pipeline

    Definition: The Continuous Delivery Pipeline (CDP) represents the workflows, activities, and automation needed to guide new functionality from ideation to an on-demand release of value.

    Figure 1 illustrates the pipeline’s four aspects: Continuous Exploration (CE), Continuous Integration (CI), Continuous Deployment (CD), and Release on Demand.

    Figure 1. The SAFe Continuous Delivery Pipeline
    Figure 1. The SAFe Continuous Delivery Pipeline

    The Continuous Delivery Pipeline (CDP) is a significant element of the Agile Product Delivery competency. Each Agile Release Train (ART) builds and maintains, or shares, a pipeline with the assets and technologies needed to deliver solution value as independently as possible. The first three elements of the CDP (CE, CI, and CD) work together to support the delivery of small batches of new functionality, which are then released to fulfill market demand.

    Details

    Building and maintaining a CDP allows each ART to deliver new functionality to users far more frequently than traditional processes. For some, continuous may mean daily or even releasing multiple times per day. For others, it may mean weekly or monthly releases—whatever satisfies market demands and the goals of the enterprise.

    Legacy practices often cause ARTs to make solution changes in large monolithic chunks. However, this does not usually need to be an all-or-nothing approach. For example, a satellite system comprises a manufactured orbital object, a terrestrial station, and a web farm that feeds the acquired data to end users. Some components may be released daily—perhaps the web farm functionality or satellite software. Other elements, like the hardware components, can only be done once every launch cycle.

    Decoupling the web farm functionality from the physical launch eliminates the need for a monolithic release. It also increases Business Agility by allowing the delivery of solution components in response to frequent market changes.

    The Four Aspects of the Continuous Delivery Pipeline

    The SAFe CDP contains four aspects: continuous exploration, continuous integration, continuous deployment, and release on demand. The CDP enables organizations to map their current pipeline into a new structure and use relentless improvement to deliver value to customers. Internal feedback loops often identify process improvements, while external feedback identifies solution improvements. The improvements collectively create synergy in ensuring the enterprise is ‘building the right thing, the right way’ and frequently delivering value to the market. The paragraphs below describe each aspect.

    • Continuous Exploration (CE) focuses on creating alignment on what needs to be built. In CE, design thinking ensures the enterprise understands the market problem or customer need and the solution required to meet that need. It starts with a hypothesis of something that will provide value to customers. Ideas are then analyzed and further researched, leading to the understanding and convergence of the requirements for a Minimum Viable Product (MVP) or Minimum Marketable Feature (MMF). These feed the solution space for exploring how existing architectures and solutions can or should, be modified. Finally, convergence occurs by understanding which Capabilities and Features, if implemented, are likely to meet customer and market needs. Collectively, these are defined and prioritized in the ART Backlog.
    • Continuous Integration (CI) focuses on taking features from the ART backlog and implementing them. In CI, the application of design thinking tools in the problem space focuses on the refinement of features (for example, designing a user story map), which may motivate more research and the use of solution space tools (such as user feedback on a paper prototype). After specific features are clearly understood, Agile Teams implement them. Completed work is committed to version control, built and integrated, and tested end-to-end before being validated in a staging environment.
    • Continuous Deployment (CD) takes the changes from the staging environment and deploys them to production. At that point, they’re verified and monitored to ensure they are working correctly. This step makes the features available in production, where the business determines the appropriate time to release them to customers. This aspect allows the organization to respond, rollback, or fix forward when necessary.
    • Release on Demand (RoD) is the ability to make value available to customers together or in a staggered fashion-based on market and business needs. This approach allows the business to release when market timing is optimal and carefully controls the risk associated with each release. Release on demand also encompasses critical pipeline activities that preserve solutions’ stability and enduring value long after release.

    Although described sequentially, the pipeline isn’t strictly linear. Instead, it’s a learning cycle that allows teams to establish one or more hypotheses, build a solution to test each one, and learn from that work, as Figure 2 illustrates.

    Figure 2. The CDP fosters continuous learning and value delivery
    Figure 2. The CDP fosters continuous learning and value delivery

    Although a single feature flows through the Value Stream sequentially, the teams work through all aspects in parallel. That means that ARTs and Solution Trains, throughout every PI and every iteration in the PI, continuously:

    • Explore user value
    • Integrate and demo value
    • Continuously deploy to production
    • Release value whenever the business needs it

    Start by Mapping the Current Workflow

    Value Stream Mapping

    Successful enterprises already have a delivery pipeline—otherwise, they wouldn’t be able to release any value. But too often, they are not fully automated, contain significant delays, and require tedious and error-prone manual intervention. These factors cause organizations to delay releases, increasing their size and scope (“We’ll release when it is big enough”). This approach is the opposite of the SAFe Principle #6, which makes value flow without interruption.

    The first step to improving value flow is mapping the current pipeline. Figure 3 illustrates the flow of value through one enterprise’s CDP, focusing initially on new feature development. Over time, this map would be extended to capture any change to the system, from new features to maintenance to architectural improvements.

    Figure 3. A map of one company’s delivery pipeline
    Figure 3. A map of one company’s delivery pipeline

    Once the current pipeline has been mapped, metrics can be collected and recorded on the value stream map to understand where delays occur. These metrics enable the ART to identify opportunities for improvement (such as eliminating delays or reducing rework). Four primary metrics [1] are used (Figure 4):

    • Active time is the time it takes to complete work in any given step. For example, in Figure 4, the ‘Design’ step takes four hours.
    • Wait time is the time between steps when no work is happening. Figure 4 shows that the features accepted by the Product Manager are delayed a staggering 696 hours before being deployed to staging. Locating and eliminating excessive wait time is critical to improving the flow of value.
    • Percent complete and accurate (%C&A) represents the percentage of work that the next step can process without needing rework. Often, delays are caused by poor quality in the upstream (prior) steps. The %C&A metric helps identify steps where poor quality might be causing delays in value delivery. Figure 4 indicates that 20% of the time, the work moving from the ‘Design’ step to the ‘Code’ step needs to be reworked. Improving the %C&A metric is also essential to improving the flow of value. All individual %C&A values are summed to produce the total %C&A, which captures the likelihood that an item will pass through the entire workflow without rework. With a cumulative total %C&A of 35%, this value stream requires most features to be reworked.
    Figure 4. Value stream map with flow metrics
    Figure 4. Value stream map with flow metrics

    Align the Current Workflow to the Continuous Delivery Pipeline

    Once the current flow is understood, it can be mapped into the SAFe CDP. Mapping helps the organization adopt a shared mental model and efficiently communicate changes and improvements. Figure 5 removes the continuous labels because, at this stage, the process is unlikely to resemble an automated pipeline.

    Figure 5. SAFe’s Continuous Delivery Pipeline mapped to a value stream
    Figure 5. SAFe’s Continuous Delivery Pipeline mapped to a value stream

    Identify Opportunities for Improvement

    Teams look for the opportunity to improve the efficiency of each step, consequently reducing end-to-end flow time. This improvement includes addressing active time and quality (percent complete and accurate) at each step.

    As shown in Figure 6, the delay between steps is often the most significant initial factor. This process has two considerable delays and a substantial amount of rework in the first step of the deployment process. Reducing wait time is typically the fastest way to significantly improve flow time. Another high-priority area to improve is any step with a low %C&A. Reducing rework enables the ART to focus on creating value rather than correcting defects. Subsequent opportunities for improvement focus on reducing the batch size and applying the DevOps practices identified in each of the specific articles describing the continuous delivery pipeline. The Team Flow and ART Flow articles provide further guidance on how to make value flow without interruption (Principle #6).

    Figure 6. Value stream maps reveal major delivery bottlenecks
    Figure 6. Value stream maps reveal major delivery bottlenecks

    Tracking Continuous Delivery

    Continuous delivery is an extensive process. Indeed, it may be the most vital capability of every ART and Solution Train. Product Management and its stakeholders should visualize and track ongoing work, even though a significant portion of the development is automated. The ART Kanban facilitates the flow of features through the CDP. Work in Process (WIP) limits help improve throughput and identify and address bottlenecks. Figure 7 illustrates a typical ART Kanban, example policies, and WIP limits governing each state.

    Figure 7. An example ART Kanban board
    Figure 7. An example ART Kanban board

    The Kanban systems consist of a series of states, each of which is summarized below:

    • Funnel – This is the capture state for all new features or enhancement of existing system features.
    • Analyzing – Features that best align with the vision are pulled into the analyzing step for further exploration. Here they’re refined with critical attributes, including the business benefit hypothesis and acceptance criteria.
    • Ready – After analysis, higher-priority features move to the backlog, where they’re ranked.
    • Implementing – At every PI boundary, top features from the ART backlog are pulled into the implementing stage, where they’re developed and integrated into the system baseline.
    • Validating on staging – Features ready for feedback get pulled into this step to be integrated with the rest of the system in a staging environment and then tested and validated.
    • Deploying to production – When capacity is available, features are deployed into the production environment, where they await release.
    • Releasing – Features are released once a sufficient amount of value has been created to meet market demands and the benefit hypothesis is evaluated.
    • Done – When the hypothesis has been satisfied, no further work on the feature is necessary, and it moves to the done column.

    Enable the Continuous Delivery Pipeline with DevOps

    Building, maintaining, and optimizing a continuous delivery pipeline requires specialized skills and tools throughout the entire value stream. Because this type of delivery system calls for the rapid delivery of complex solutions with very short learning loops and high degrees of cross-functional collaboration, DevOps methods are ideally suited to enable it. In other words, continuous delivery pipelines are best implemented with DevOps, as illustrated in Figure 8.

    Figure 8. DevOps enables the CDP
    Figure 8. DevOps enables the CDP

    SAFe’s CALMR approach to DevOps is a mindset that guides continuous value delivery by improving Culture, Automation, Lean Flow, Measurement, and Recovery. DevOps technical skills, practices, and tooling are grouped into practice domains, represented by the model’s inner loops. The two outer loops represent the four aspects of the CDP, each of which has four activities.

    Download the free DevOps Health Radar assessment here. AgilityHealth also offers an online version of this assessment.

    Learn More

    [1] Martin, Karen. Value Stream Mapping: How to Visualize Work and Align Leadership for Organizational Transformation. McGraw-Hill Education, 2013.

    [2] Kim, Gene. The Phoenix Project: A Novel about IT, DevOps, and Helping Your Business Win. IT Revolution Press, 2013.

    [3] Kim, Gene, Jez Humble, Patrick Debois, and John Willis. The DevOps Handbook: How to Create World-Class Agility, Reliability, and Security in Technology Organizations. IT Revolution Press, 2016.

    Last update: 26 September 2023