A developer needs to update an important backend service used for processing data. This service is critical, and any downtime must be avoided. The developer wants to transition the traffic gradually to observe the new behavior without disrupting the current users. What deployment technique should be applied to achieve this zero-downtime update and facilitate observability and quick rollback if issues arise?
Directly overwrite the current service code with the new update, relying on the underlying system to manage the transition seamlessly.
Configure a brand-new endpoint in the managed API service, directing a copy of incoming requests to the updated backend version for testing.
Introduce a new version of the service and leverage the progressive traffic shifting feature to incrementally route user requests, monitoring the update's stability before redirecting all traffic.
Modify the service's environment properties to point to the new version, which will transparently divert the traffic.