You support a web application that is hosted on Compute Engine. The application provides a booking service for thousands of users. Shortly after the release of a new feature, your monitoring dashboard

The problem statement explicitly links the latency issue to 'shortly after the release of a new feature.' In such a scenario, where a new deployment causes severe service degradation (all users experiencing latency), the quickest way to mitigate user impact and restore service health is to roll back to the previously working version of the application. This is a primary SRE incident response strategy.

Reviewing Stackdriver monitoring is important for understanding the problem, but as a *first* mitigation step when the cause is clearly linked to a new release, rolling back takes precedence to minimize user impact.

Upsizing VMs might temporarily alleviate symptoms if the issue is a resource bottleneck, but it doesn't address the root cause if the new feature itself introduced a performance regression or bug.

Deploying a *new* release immediately without a clear understanding of the fix is extremely risky and could exacerbate the problem or introduce new problems, going against safe deployment practices.