Question

A cloud operations team monitors a fleet of Linux virtual machines running a memory-intensive analytics application. The provider's default monitoring dashboard shows CPU, disk, and network metrics, but the team keeps missing out-of-memory crashes because memory usage is not visible. What is the MOST appropriate way to gain visibility into memory consumption?

Accepted Answer

Install a monitoring agent on each instance to collect and publish memory utilization as a custom metric

Answer

Increase the sampling frequency of the existing default CPU and disk metrics

Answer

Enable the provider's flow logs to capture memory pressure events from the hypervisor

Answer

Resize the instances to a larger memory tier so out-of-memory crashes stop occurring