How does your CI/CD strategy differ between monoliths and microservices?

WHAT THIS TESTS: This question tests whether you understand release coupling and organizational scalability. Interviewers want to see that you recognize a monolith's single artifact model creates coordination overhead and release trains, while microservices shift complexity toward distributed pipeline ownership, trust verification, and blast radius isolation. They are listening for architectural reasoning, not just tool names.

A GOOD ANSWER COVERS: First, monolith CI/CD: a single build pipeline produces one deployable artifact, so every change must integrate before release. This creates release trains where Team B's bug can block Team A's feature. Quality gates focus on end-to-end integration tests, and rollbacks mean reverting the entire application. Second, microservice CI/CD: each team owns its pipeline and can release independently. The answer should mention side-by-side deployments, quality gates per service, and trust mechanisms like signed container images, SBOM attestations, and vulnerability scanning. Third, trade-offs: monoliths have simpler dependency management but slower velocity and higher coordination cost; microservices enable faster cycles but risk pipeline sprawl, fragmented deployment knowledge, and harder distributed integration testing. Fourth, access control: microservices pipelines should use federated short-lived credentials rather than long-lived secrets.

COMMON WRONG ANSWERS: A red flag is proposing one shared pipeline for all microservices, which recreates monolithic coupling and destroys team autonomy. Another is ignoring rollback strategy for monoliths or pretending microservices eliminate integration testing entirely. Candidates who only list tools without explaining coupling, blast radius, or release train dynamics miss the architectural point.

LIKELY FOLLOW-UPS: How do you handle integration testing when each microservice deploys independently? How do you secure pipeline credentials across dozens of repositories? What is your rollback strategy when a monolith release fails versus a canary failure in one microservice? How do you prevent pipeline sprawl and standardize quality gates without centralizing ownership?

ONE CONCRETE EXAMPLE: In a monolith, a critical bug fix in a payment module requires rebuilding the entire application, rerunning the full integration suite, and coordinating a production deployment window because the artifact is indivisible. In a microservices system, the payment service team rebuilds only their container, the pipeline signs the image and attaches an SBOM, it deploys side-by-side with the previous version behind a feature flag, and if error rates spike, only that service rolls back without affecting inventory or shipping services.

Source: learn.microsoft.com (CI/CD for Microservices - Azure Architecture Center)