Implement change - Reliability Pillar

Implement change

Controlled changes are necessary to deploy new functionality and to ensure that the workloads and the operating environment are running known, properly patched software. If these changes are uncontrolled, then it makes it difficult to predict the effect of these changes, or to address issues that arise because of them.

Additional deployment patterns to minimize risk

Feature flags (also known as feature toggles) are configuration options on an application. You can deploy the software with a feature turned off, so that your customers don’t see the feature. You can then turn on the feature, as you’d do for a canary deployment, or you can set the change pace to 100% to see the effect. If the deployment has problems, you can simply turn the feature back off without rolling back.

Fault isolated zonal deployment: One of the most important rules AWS has established for its own deployments is to avoid touching multiple Availability Zones within a Region at the same time. This is critical to ensuring that Availability Zones are independent for purposes of our availability calculations. We recommend that you use similar considerations in your deployments.

Operational Readiness Reviews (ORRs)

AWS finds it useful to perform operational readiness reviews that evaluate the completeness of the testing, ability to monitor, and importantly, the ability to audit the application's performance to its SLAs and provide data in the event of an interruption or other operational anomaly. A formal ORR is conducted prior to initial production deployment. AWS will repeat ORRs periodically (once per year, or before critical performance periods) to ensure that there has not been drift from operational expectations. For more information on operational readiness, see the Operational Excellence pillar of the AWS Well-Architected Framework.

Best practices