Overview
Direct Answer
A Rolling Update is a deployment strategy that incrementally replaces running instances of an application with a new version whilst maintaining service availability. It eliminates the need for complete service downtime by distributing the upgrade across a sequence of controlled replacements.
How It Works
The mechanism removes a small subset of running instances from load balancers, upgrades them to the new version, and returns them to the pool before proceeding to the next batch. This continues iteratively until all instances run the new version. The process is governed by parameters such as the number of instances replaced per cycle and the health-check interval between cycles.
Why It Matters
Organisations adopt this approach to minimise user-facing disruption during deployments whilst maintaining predictable capacity and response times. The strategy reduces operational risk by enabling rapid rollback if issues are detected, and allows teams to validate new releases against live traffic patterns incrementally.
Common Applications
Rolling updates are standard in containerised environments managed by orchestration platforms, microservices architectures, and cloud-native applications where horizontal scaling is deployed. Common scenarios include updating web service fleets, database connection pools, and load-balanced API gateways.
Key Considerations
The strategy requires backward compatibility between versions during the transition window and careful management of database schema changes. Performance validation and health checks must be sufficiently robust to detect failures before all instances are replaced.
Cross-References(1)
More in DevOps & Infrastructure
Prometheus
ObservabilityAn open-source monitoring and alerting toolkit designed for reliability and scalability in cloud-native environments.
Vertical Scaling
CI/CDIncreasing the resources (CPU, RAM, storage) of an existing machine to handle more load.
Helm
Containers & OrchestrationA package manager for Kubernetes that simplifies the deployment and management of applications using charts.
Alerting
ObservabilityAutomated notifications triggered when system metrics or conditions exceed predefined thresholds.
Puppet
Infrastructure as CodeA configuration management tool that automates the provisioning and management of infrastructure.
Graceful Degradation
CI/CDA design approach where a system continues to operate with reduced functionality when components fail.
Site Reliability Engineering
Site ReliabilityA discipline applying software engineering principles to infrastructure and operations to create scalable, reliable systems.
Observability
ObservabilityThe ability to understand a system's internal state from its external outputs, encompassing metrics, logs, and traces.