In a canary deployment, what metric would you monitor to decide whether to increase traffic to the new version?

In a canary deployment, what metric would you monitor to decide whether to increase traffic to the new version?

• A. Health indicators such as error rate and latency
• B. Number of active sessions
• C. CPU usage of old version
• D. Disk space on primary node

Answer: (A)

In a canary deployment, you expand traffic to the new version only as its health under load stays acceptable. The best metric to monitor is health indicators such as error rate and latency because they directly reflect how users are experiencing the service—whether requests are being answered correctly and quickly. If these metrics stay within acceptable bounds, you can safely ramp up. If they worsen, you should pause or roll back. Monitoring the number of active sessions doesn’t reveal whether failures are occurring; CPU usage of the old version doesn’t show how the new version handles requests under load; and disk space on the primary node is an infrastructure concern that doesn’t indicate runtime reliability.

In a canary deployment, you expand traffic to the new version only as its health under load stays acceptable. The best metric to monitor is health indicators such as error rate and latency because they directly reflect how users are experiencing the service—whether requests are being answered correctly and quickly. If these metrics stay within acceptable bounds, you can safely ramp up. If they worsen, you should pause or roll back. Monitoring the number of active sessions doesn’t reveal whether failures are occurring; CPU usage of the old version doesn’t show how the new version handles requests under load; and disk space on the primary node is an infrastructure concern that doesn’t indicate runtime reliability.