Key Findings from the 2017 Puppet State of DevOps Report

The Puppet State of DevOps Report 2017, available at https://puppet.com/blog/2017-state-devops-report-here , shows that high‑performing IT teams are deploying more frequently and recovering faster, with a strong focus on automation, loosely coupled architectures, and lean product management.

The annual survey reached about 3,200 respondents worldwide, including executives, developers, and IT professionals. Respondents working on DevOps teams grew from 16% to 27% over three years, with most participants from organizations of 100‑499, 500‑1,999, and 10k+ employees across DevOps, IT Ops/Infrastructure, and Development/Engineering roles.

Key performance metrics measured were deployment frequency, lead time for changes, mean time to recover (MTTR), and change failure rate.

High‑performing teams improved on all metrics, achieving 46‑times more frequent deployments and a 96‑times faster MTTR compared to low‑performing teams, which also showed year‑over‑year improvements.

Automation practices rose significantly, with 28% more configuration‑management automation and 26% more deployment automation among high‑performers.

Continuous delivery capability was assessed by on‑demand production deployments and fast feedback loops, driven by comprehensive version control, CI, trunk‑based deployments, security integration, and test/deployment automation. Loose coupling of services and teams was identified as a critical architectural factor.

Non‑technical drivers included lean product management, defined as splitting work into small batches with visible workflow, gathering and acting on customer feedback, and empowering development teams to modify specifications without additional approvals.

Leadership alone is insufficient; success also depends on suitable architecture and robust technical practices. The report employed a Structured Equation Model (SEM) to predict relationships among constructs relevant to software delivery.

The accompanying figure illustrates statistically significant relationships between these factors.

All images are courtesy of the 2017 State of DevOps Report.