DevSecOps Jun - 2022.

Building & Managing the DevSecOps Practice Using Azure DevOps

How to embed security into every phase of the software delivery lifecycle at enterprise scale — covering infrastructure automation, pipeline templatization, shift-left security, and zero-trust at India's largest private-sector conglomerate.

01. Introduction

In today's rapidly evolving IT landscape, security and compliance are no longer an afterthought — they are a prerequisite. Traditional development and operations models have proven insufficient against the increasing velocity of cyber threats and the growing weight of regulatory requirements across industries.

This case study documents the establishment of a comprehensive DevSecOps practice at India's largest private-sector conglomerate using Azure DevOps as the delivery backbone. The program unified people, process, and tooling to achieve continuous, secure delivery across a heterogeneous technology portfolio spanning manufacturing, retail, energy, and digital services.

"DevSecOps is the union of people, process and products to enable continuous delivery of value to your end users." — Donovan Brown, Microsoft

02. Vision & Objectives

The core vision was to integrate security into every phase of the software development lifecycle (SDLC) while optimising pipeline performance and maintaining developer accessibility. Security could not be a gate at the end — it had to be woven into every commit, build, and deployment.

The programme was anchored on five measurable objectives:

03. The DevSecOps Pipeline: Security at Every Stage

The foundation of the practice is a secure, templatized CI/CD pipeline where every stage has a security responsibility. The diagram below shows the integrated pipeline flow and the tools applied at each phase.

Plan Threat Model Code Pre-commit hooks Build SAST · SCA Test DAST · Perf · API Approval Gate Dev·Stage·UAT·Prod Deploy IaC · Ansible Monitor Zero Trust · IAM SonarQube · HP Fortify Black Duck · Nexus HP WebInspect · JMeter Robot · SoapUI · Postman Continuous feedback loop Single Build · Multiple Deployment · Templatized per Technology Stack
Figure 1 — DevSecOps pipeline with integrated security tooling at every stage. A single build artifact flows through environment-specific approval gates before deployment.

04. Approach: Security Built Into the Factory

The programme did not layer security onto an existing pipeline — it redesigned the delivery factory from the ground up. Six interconnected design principles governed the architecture.

🏗️
Infrastructure as Code (IaC)

Environment provisioning was fully automated using Terraform. Every infrastructure component — clusters, namespaces, network policies, secrets stores — was version-controlled, reviewed, and enforced through the same pipeline that shipped application code. Drift was eliminated by design.

End-to-End Repository Automation

Onboarding a new application triggered automated creation of a repository with a standard folder structure and sample code, build and release pipelines, approval groups, artifact storage, and authorisation provisioning across InfoSec tools — all without manual intervention from the platform team.

🔒
CI/CD Security Integration

Static Application Security Testing (SAST) via SonarQube and HP Fortify, Dynamic Application Security Testing (DAST) via HP WebInspect, and Software Composition Analysis (SCA) via Black Duck were embedded as mandatory pipeline stages. A build could not proceed unless quality gates were cleared.

🎛️
Deployment Flexibility

Technical Leads could define and override deployment parameters — memory, CPU, HPA scaling, volume mounts, port configuration, Kubernetes cluster and namespace, replica count, artifact repository, and API gateway routing — without touching pipeline code. This put resource governance in the hands of the people closest to the service.

Approval Matrix

A structured approval matrix governed promotion across environments — Development, Staging, UAT/Replica, and Production. Each promotion required sign-off from the appropriate team, with escalation paths and time-bound gates to prevent bottlenecks.

📐
Pipeline Templatization

Build and release pipelines were templatized for every technology in the portfolio: Node.js, Spring Boot, Go, Python, Angular, React, .NET Core, Flutter, relational databases, Kubernetes resource kinds (ConfigMap, Secrets, Gateway Service, Virtual Service, HPA, CronJob), and SAP Transport. Teams received a working, secure pipeline on day one, not a blank canvas.

05. Challenges & Solutions

Three structural challenges shaped the design of the practice. Each required a targeted engineering response rather than a generic best-practice overlay.

Legacy Systems Integration

Challenge

Older applications in the manufacturing division lacked modern security frameworks. They could not be instrumented by standard SAST agents or retrofitted with identity libraries without extensive rework.

Solution

An API-based security layer was introduced using WSO2 API Management and Web Application Firewall (WAF). This provided perimeter-level threat detection, rate limiting, and policy enforcement without modifying legacy application code.

Performance Optimisation vs. Security Overhead

Challenge

Adding SAST, DAST, SCA, and automated testing to every pipeline run added significant build time, threatening developer productivity and deployment cadence commitments.

Solution

Scan results were cached intelligently between builds, and security test stages were parallelised where dependency allowed. Incremental scans replaced full scans on short-lived feature branches, reserving comprehensive scans for main and release branches.

Securing a Multi-Cloud & Hybrid Estate

Challenge

The organisation operated a mix of on-premise data centres, Azure, GCP, and AWS infrastructure. No single commercial tool covered the full estate, and multiple vendor dashboards created blind spots.

Solution

A combination of in-house integrations and carefully selected COTS products was assembled to create a unified security management plane. Policy definitions, vulnerability findings, and compliance status were aggregated into a single view regardless of infrastructure provider.

06. Solutions & Best Practices

Shift-Left Security

Security was moved as far left in the SDLC as possible. Pre-commit hooks enforced code standards and flagged obvious vulnerabilities before code left the developer's machine. Automated threat modelling ran during the design phase, surfacing potential attack surfaces before architecture decisions were locked.

The result was a measurable reduction in security findings at later pipeline stages — defects caught at commit time cost a fraction of those found in staging or production.

Zero Trust Architecture

The practice adopted Zero Trust as the governing identity model. "Never trust, always verify" was implemented through Open Source IAM tooling enforcing Conditional Access policies — every service-to-service call, every human operator action, and every CI/CD agent interaction required verified identity and least-privilege authorisation.

This eliminated the implicit trust that had previously existed within the corporate network perimeter and made lateral movement significantly harder for any threat actor who gained initial access.

07. Technology Stack

The practice was built on a curated set of tools, each selected for a specific security or delivery capability within the pipeline.

Pipeline & Orchestration

Azure DevOps Ansible Terraform

Static Analysis (SAST)

SonarQube HP Fortify

Dynamic Analysis (DAST)

HP WebInspect

Composition Analysis (SCA)

Black Duck Nexus

Functional & API Testing

Robot Framework SoapUI Postman MochaChai Cucumber Jasmine Karate

Performance Testing

JMeter

Build & Scripting

Java Python Ant Groovy

08. Conclusion

Building a DevSecOps practice at this scale required more than tooling — it required a cultural and architectural shift in how the organisation thought about security's role in delivery. By making security an engineering discipline embedded in every pipeline stage, every repository template, and every deployment decision, the organisation achieved what traditional security programmes rarely do: compliance and speed operating in parallel rather than in opposition.

The practice established a secure, scalable, and high-performing software delivery pipeline. Development teams gained a working, policy-compliant pipeline on day one. Security teams gained real-time visibility across the entire application estate. Operations teams gained confidence that what went to production had passed a consistent, automated bar.

The outcome was faster time-to-market, measurably reduced vulnerability exposure, continuous compliance evidence, and a delivery system positioned to absorb whatever new threats and regulatory requirements the next phase of enterprise growth demands.

Security is not a gate you place at the end of the pipeline. It is the material you build the pipeline from.
Back To Blog