Datadog

Datadog brings infrastructure and application monitoring into a hosted service where teams can explore measurements and investigate incidents. Its Kubernetes agents collect information from the cluster and send the enabled signals to your Datadog account. It is useful when you want cluster activity alongside monitoring from other systems. Metrics, logs and traces are separate choices, so install the collection features you need rather than assuming every Agent deployment gathers the same data.

Deployment and operating notes

Datadog currently documents installing its Kubernetes Agent using the Datadog Operator, Helm or manifests, with Agent and Cluster Agent compatibility guidance. The historical stable chart and its optional kube-state-metrics dependency do not describe every current collection mode. Infrastructure metrics, logs, traces and security features have separate configuration and operational implications.

Inventory the existing site endpoint, cluster name, tags, secrets, autodiscovery rules and enabled features before moving packaging. Compare host mounts, capabilities and network access with the chosen distribution’s restrictions. During migration, verify one known metric, log and trace only for the features you intend to collect, and look for duplicate agents or duplicate cluster-state ingestion. Check labels and cardinality before recreating old dashboards because changed dimensions can affect both alerts and usage. Keep API keys in a managed secret path and preserve previous values for rollback. A ready Agent pod does not confirm that data reaches the correct account or that billing-relevant collection is constrained.