A self-paced, practical course for developers, DevOps and platform engineers who need to ship changes without taking services offline. Seven lessons cover the causes of deployment downtime, rolling, blue-green and canary patterns, load balancing and health checks, backward-compatible database migrations, feature flags and dark launches, automated verification and rollback, and running the release process day to day. Each lesson ends with an exercise you can apply to a real service.
What you will learn
- Identify the causes of downtime during deployments
- Choose between rolling, blue-green and canary deployments
- Configure health checks, connection draining and graceful shutdown
- Run backward-compatible database migrations with expand and contract
- Use feature flags to separate deploy from release
- Automate verification and rollback
- Operate a repeatable zero-downtime release process
Lessons
- Why Deployments Cause Downtime
- Rolling, Blue-Green and Canary Deployments
- Health Checks, Draining and Graceful Shutdown
- Backward-Compatible Database Migrations
- Feature Flags and Dark Launches
- Automated Verification and Rollback
- Operating the Release Process
Self-paced course from Tech Skills Library. Educational content only; it is not an accredited qualification, licence or credential.
Lessons
Why Deployments Cause Downtime
Rolling, Blue-Green and Canary Deployments
Three main patterns support zero-downtime deployment. Each trades off cost, complexity and risk. Rolling deployments Replace instances in batches. At each step, new instances start, pass health checks and receive traffic before old instances are removed. Requires both versions to run side by side safely. Low extra cost, but rollback means rolling back through the […]
Health Checks, Draining and Graceful Shutdown
Traffic must only reach instances that are ready, and instances must finish their work before stopping. Readiness and liveness Readiness checks indicate whether an instance can serve traffic. Fail readiness until start-up tasks, such as loading configuration and warming caches, are complete. Liveness checks indicate whether an instance is functioning. Failing liveness triggers a restart. […]
Backward-Compatible Database Migrations
Database changes are the most common obstacle to zero-downtime deployment, because old and new application versions share the same database during rollout. Expand and contract Expand: add new structures without removing old ones, such as adding a nullable column. Migrate: deploy code that writes to both old and new structures and backfill data. Switch: deploy […]
Feature Flags and Dark Launches
Feature flags separate deploying code from releasing features. This reduces deployment risk and enables gradual rollout. How flags work Code checks a flag to decide whether to execute new behaviour. Flags can be toggled without deploying, for all users or for targeted groups. Uses Release features gradually. Turn off problematic features instantly. Test in production […]
Automated Verification and Rollback
A zero-downtime system needs automated checks that confirm a release is healthy and roll back quickly when it is not. What to verify Error rates and status codes. Latency percentiles. Saturation such as CPU and memory. Business metrics like successful checkouts or logins. Synthetic transactions for critical journeys. Automated gates Define thresholds that must hold […]
Operating the Release Process
Zero-downtime deployment becomes routine when supported by a clear, repeatable process. Release checklist Change reviewed and tested in the pipeline. Database migration plan confirmed backward compatible. Feature flags configured. Monitoring dashboards ready. Rollback plan confirmed. Stakeholders informed where needed. Deploy during working hours When deployments are safe, deploy when the team is available to respond, […]