What monitoring technique should you implement for alerting when your application is unavailable?

The most suitable monitoring technique for alerting when your application is unavailable is Stackdriver uptime checks. This service allows you to configure checks based on URL endpoints to monitor the availability of your application. It periodically makes HTTP or TCP requests to your specified endpoints and determines if they respond successfully. If the requests fail consistently over a set threshold (e.g., failures over a specific duration), Stackdriver will send you alerts indicating that your application is down. This proactive monitoring helps ensure that you are promptly informed of any outages, allowing for rapid response and remediation.

In contrast, other options, while valuable for monitoring and health checks, do not specifically target the availability aspect in the way that uptime checks do. Smoke tests, for instance, validate basic functionality rather than availability, and cloud load balancing health checks are primarily for assessing the state of backend instances rather than the application itself. Managed instance group health checks are geared towards ensuring that instances within a managed group are healthy, but they do not capture the broader application availability context.