Managing cloud costs can be challenging as Kubernetes fleets scale. Grafana Cloud has introduced a cost-monitoring feature within Kubernetes Monitoring to address this issue. In particular, Grafana Cloud’s Kubernetes Monitoring offers ML predictions for CPU and memory usage. It can also identify CPU outlier pods, allowing users to anticipate potential resource issues before they impact mission-critical infrastructure or budgets.
To generate forecasts for CPU and memory usage is necessary to specify a source query (representing the time series to be modeled) and configure the machine learning model. Subsequently, the system autonomously trains the model in the background.