AZ-400 is current, but its 2026 objectives matter
AZ-400 is Microsoft’s current Designing and Implementing Microsoft DevOps Solutions exam. It is associated with the DevOps Engineer Expert certification, and Microsoft lists no retirement date on the active exam page. The English-language skills measured changed on July 27, 2026. Older guides may still describe relevant DevOps concepts, but they can omit new examples of GitHub security, workload identity or pipeline configuration. For an exam decision, distinguish Microsoft’s published skills outline from a third-party article’s interpretation, because the issuer determines what can be assessed and periodically updates the details.
The five currently published domains are design and implement processes and communications (10–15 percent), design and implement a source control strategy (10–15 percent), design and implement build and release pipelines (50–55 percent), develop a security and compliance plan (10–15 percent), and implement an instrumentation strategy (5–10 percent). These percentages are domain ranges rather than quotas for practice questions. A scenario about a failed deployment might combine several of them. More than half the blueprint focuses on delivery automation, but a candidate who ignores identity or observability can still choose an unsafe solution.
The intended audience spans developers and infrastructure administrators who already understand at least one side of Azure delivery deeply and collaborate across engineering, operations and security. Microsoft also expects familiarity with both GitHub and Azure DevOps solutions. That does not mean every candidate must use both products for every stage; it means you should recognize their core workflows and select the right control under a given scenario. Exam-specific details and current administrative policies should always be checked on the Microsoft certification page rather than inferred from dated screenshots.
A successful AZ-400 exam contributes to the DevOps Engineer Expert credential, but the credential also requires a qualifying Microsoft Azure Administrator Associate or Azure Developer Associate certification. The latter pathway’s exam availability can change independently; candidates should check the currently valid prerequisite credential status. Passing AZ-400 should not be described as immediately awarding an expert certification to someone with no qualifying prerequisite. A conceptual introduction to AZ-400 and DevOps fundamentals is a useful starting point, but its overview cannot substitute for the active issuer objective list.
Processes, communications and source control
The processes-and-communications domain concerns traceable work, meaningful metrics and collaboration. Microsoft lists GitHub Flow, feedback cycles, GitHub Issues and projects, Azure Boards integration, dashboards, queries, release notes, documentation and webhooks. Understanding the words is not enough: a plausible question may present a team that can deploy quickly but cannot connect a production incident to its originating requirement. The stronger control is one that preserves a path from work item through commit, pull request, build and release. An isolated dashboard of ticket counts cannot establish deployment accountability.
Metrics must match the decision they are intended to support. Cycle time and lead time can expose bottlenecks, while mean time to recovery and deployment failure rates can reveal operational instability. A team can make a metric look better by splitting work differently or suppressing incidents, so the interpretation needs context. Azure Boards and GitHub integrations can organize evidence when a team actually uses the same source IDs and release metadata. The value of Azure Boards lies in the relationship between delivery requests, acceptance and incidents, not merely in naming a backlog tool.
Source-control strategy includes trunk-based, feature and release branching, pull-request policy, merge restrictions, repository permissions, tags, repository recovery and safe removal of unwanted content. The guide also explicitly references large-file strategies such as Git LFS and approaches for scaling repositories. A case about a binary-heavy monorepository may require different controls from a small web application. A case about a leaked credential requires revocation as well as repository-history remediation. Removing the secret only from the latest commit does not undo exposure.
Choose a branching strategy from release cadence, support requirements and integration risk. Long-lived feature branches can create costly merge conflicts, but a supported customer release may need a separate maintenance line. Required reviews and status checks protect a branch, yet an emergency hotfix also needs a clear approval and rollback process. Answering a source-control question therefore requires balancing collaboration, technical constraints and operational recovery rather than treating one Git command as universally correct.
Why the build-and-release domain occupies 50–55 percent
Package management covers Azure Artifacts, GitHub Packages, feeds, upstream dependencies, versions and retention. A stable package version or artifact digest connects a tested change to a deployed product. A release that silently rebuilds from the moving tip of a branch can differ from what was approved. A candidate should be able to trace commit, build output, dependency versions and deployment environment. Semantic versioning and date-based versioning both appear in the guide, but neither can rescue a workflow that consumes an unidentifiable artifact.
Pipeline testing includes local, unit, integration and load tests, test agents, coverage reporting and release gates. The goal is to choose checks that reveal relevant defects without turning the build into an uncontrolled collection of redundant work. A high code-coverage percentage is not evidence that a customer transaction succeeds. If an integration test fails only when a service credential expires, the engineer must distinguish test design from infrastructure and identity problems. The subject of Azure Pipelines becomes clearer when job results and stage gates are interpreted as evidence, not just status colors.
Pipeline implementation includes GitHub Actions and Azure Pipelines; YAML; trigger rules; reusable templates; job ordering and parallelism; hosted and self-hosted agents; and environment approvals. The exam may ask which architecture fits a private-network deployment or how a shared runner pool affects cost, isolation and reliability. A reusable template can improve consistency across teams but may also spread a faulty change widely. Versioning, ownership and exception handling make shared automation safer. Classic-to-YAML pipeline migration and artifact retention are within the maintenance scope, not optional historical trivia.
The deployment objectives include progressive techniques such as canary, blue/green, rings, feature flags and A/B tests, along with containers, binary releases, scripts and database tasks. Candidates should choose an exposure strategy based on traffic behavior, application health and rollback compatibility. A feature flag can disable behavior without restoring code, while a blue/green switch can be dangerous when the earlier version cannot understand a new database schema. Infrastructure as code, including Bicep, ARM, desired-state configuration and Azure Deployment Environments, is assessed within this large pipeline domain. Revision control must extend to the environment itself.
Security, compliance and instrumentation are cross-cutting
The security-and-compliance domain addresses Azure DevOps and GitHub permissions, service connections, GitHub Apps, tokens, Microsoft Entra service principals, managed identities, workload identity federation and secret handling. Selecting the smallest valid privilege is more important than memorizing which credential type has the most features. A deployment identity that can modify unrelated production subscriptions is unnecessarily powerful, even if it makes the workflow easier to configure. Distinguish who can trigger a release, which role performs it and which resources that role can actually modify.
Azure Key Vault, secure files and federated authentication can help prevent sensitive data from appearing in repository content or pipeline logs. Scanning adds checks for code vulnerabilities, dependency risk, exposed secrets, container images and license issues. Microsoft’s current outline includes GitHub Advanced Security, its Azure DevOps counterpart, Dependabot and integration with Defender for Cloud. A scanner finding should create a triage decision with evidence; ignoring findings or granting every pipeline broad privileges is not a mature security strategy. A relevant explanation of Azure Key Vault grounds the secret-management choices.
Instrumentation covers Azure Monitor, Application Insights, VM and container insights, storage and network telemetry, GitHub and pipeline alerts, distributed tracing, and basic Kusto Query Language queries. You must understand what data is collected and how to interpret it. A healthy host does not prove the application serves the correct data. An empty logging view may indicate a broken agent rather than an absence of incidents. Connect a trace or failed operation to the deployment version and a follow-up action. The current exam gives instrumentation a smaller percentage, but it can determine whether a deployment control is safe to automate.
Observability applies to the delivery system too. Monitor pipeline failure frequency, stage duration, flaky tests, queue delays and runner stability. A build optimization that removes valuable tests can harm release quality even while improving duration. A security control that creates unbounded queue delays may need better design rather than wholesale removal. These are examples of making the same delivery workflow faster, safer and more diagnosable instead of improving one metric in isolation.
Interpret the blueprint through specific decisions
Suppose a team uses GitHub to store code, Azure Pipelines to deploy to a private test network and Azure Monitor for application telemetry. A release is blocked after a build succeeds. The correct response depends on whether an environment approval is pending, a self-hosted agent cannot connect, the service connection lacks permission or a required test failed. The Azure AZ-400 role requires diagnosing those different evidence paths, not simply knowing that all four tools exist. Begin from the failure stage, look at its logs and identify the narrowest corrective action.
Now suppose the same team releases a new API version through a canary but user requests begin failing while CPU and memory remain normal. A useful monitoring strategy observes client errors and downstream dependencies, then pauses traffic expansion. Rolling back may be safe only if the newer version has not applied incompatible data changes. The question combines deployment pattern, testing, telemetry and operational recovery. The distinction between CI and deployment automation helps explain why a green build cannot be treated as business success.
To prepare, map each official objective to either a completed lab with evidence or an explicit knowledge gap. Keep the same application and pipeline across multiple scenarios so interactions are visible. Practice one failure in source-control review, one in artifact handling, one in identity, one in deployment compatibility and one in telemetry. Ask how you would detect the failure, why the alternative remedy is unsuitable and how you would verify recovery. This is more credible than a study plan that promises passing based solely on how many hours were spent reading.
Finally, treat the official July 2026 guide as the active scope when preparing in November 2026, and verify updates before a scheduled appointment. Microsoft may assess examples beyond a list’s named products, and localized exam updates can lag the English version. An accurate article identifies the published blueprint and its limits while encouraging practical reasoning. It does not make unsupported guarantees about exact question counts, a fixed number of study weeks, automatic certification eligibility or a particular promotion outcome.