At the end of every year, the DevOps community takes stock of where software delivery practices stand and where they are headed. Heading into 2017, continuous delivery (CD) was one of the most actively discussed topics, with teams at every scale racing to shorten release cycles, reduce deployment risk, and automate more of the software pipeline.
The predictions below reflect the themes that were dominating DevOps conversations as 2016 ended. They were originally surfaced by the team at DevOps.com and remain a useful baseline for understanding how the discipline has matured.
Why Continuous Delivery Was the Central DevOps Concern in 2017
By the end of 2016, many organisations had already adopted agile development and basic CI (continuous integration) pipelines. The natural next step — automating the path from a passing build all the way to production — was where teams were struggling. Continuous delivery promised faster time-to-market, fewer manual handoffs, and more reliable releases, but realising those benefits required significant cultural and tooling changes.
Key Themes That Defined CD Predictions for 2017
1. Pipeline Automation Would Move Beyond Code
In 2016, most CD pipelines focused on building and testing application code. The emerging expectation was that infrastructure, database migrations, configuration, and security scanning would all become first-class citizens in the pipeline — a concept now widely called “pipeline as code” and embodied in tools such as Jenkinsfile, GitLab CI, and later GitHub Actions.
2. Container Adoption Would Accelerate CD Maturity
Docker had achieved mainstream awareness by 2016, and Kubernetes was gaining traction as an orchestration platform. Predictions centred on containers becoming the standard packaging format for CD pipelines, making environment parity between development, staging, and production far easier to achieve. This would, in theory, eliminate the classic “works on my machine” class of deployment failures.
3. Security Would Shift Left
The term “DevSecOps” was still relatively new in 2016, but the pressure to embed security testing earlier in the development lifecycle was growing. Predictions called for static analysis, dependency vulnerability scanning, and compliance checks to become automated pipeline stages rather than manual gates performed by a separate security team after the fact.
4. Feature Flags Would Replace Long-Lived Branches
Long-lived feature branches were identified as one of the biggest blockers to true continuous delivery. The predicted shift was toward trunk-based development combined with feature flags (also called feature toggles), allowing teams to merge code continuously while still controlling which features were visible to end users. This approach decouples deployment from release.
5. Metrics and Observability Would Drive CD Feedback Loops
Deploying frequently is only valuable if teams can quickly detect and respond to problems in production. Predictions emphasised that CD programmes would increasingly depend on robust observability — distributed tracing, structured logging, and real-time dashboards — to close the feedback loop between deployment and business outcome.
The Cultural Dimension: CD Is Not Just a Tooling Problem
A consistent thread in end-of-2016 analysis was that technology alone would not deliver continuous delivery. Organisational structure, shared ownership of the pipeline, blameless post-mortems, and executive buy-in were all identified as critical success factors. Teams that treated CD purely as a tooling exercise consistently underperformed those that also invested in culture and process change.
How These Predictions Held Up
Looking back from the present, most of these predictions proved accurate directionally. Container-based pipelines became standard. Feature flags matured into a dedicated product category. DevSecOps became an established discipline with its own toolchain. Observability grew from a niche concern into a core engineering requirement. The pace of adoption varied widely by organisation size and industry, but the trajectory was clear.
For teams still building out their CD practice, these themes remain directly relevant. The fundamentals of pipeline automation, environment consistency, early security testing, and feedback-driven iteration have not changed — only the specific tools available to implement them have evolved.
Further Reading
The original predictions article was published on DevOps.com. For teams looking to benchmark their current CD maturity, the DORA (DevOps Research and Assessment) metrics — deployment frequency, lead time for changes, change failure rate, and time to restore service — remain the industry-standard framework.