starter.wiki

Observability

Infrastructure Monitoring

Choose how host, VM, container, Kubernetes, cloud-resource, process, and system metrics are collected, retained, explored, and alerted on.

Recommendation

Route by operating ownership and telemetry estate.

Commercial full-stack suites, a managed open-standards service, and self-operated Prometheus move different parts of collection, storage, query, and reliability to the team.12

For: Platform, infrastructure, and reliability teams responsible for production resource visibility

Main trade-off

Infrastructure monitoring improves resource visibility while shifting cardinality, retention, dashboard, alert, collector, and cost control into an ongoing operating discipline.1234

Three infrastructure-metrics routes

Datadog and New Relic share a commercial full-stack route; Grafana Cloud and Prometheus separate managed and self-operated open ecosystem choices.

  1. Resource scope

    List hosts, VMs, containers, Kubernetes, cloud services, processes, network, and required integrations.12

  2. Metrics model

    Plan labels, cardinality, scrape or agent behavior, topology, aggregation, retention, and query requirements.13

  3. Operations

    Assign collector, backend, upgrade, capacity, high-availability, backup, alert, and incident ownership.12

  4. Cost and exit

    Model hosts, containers, custom metrics, ingest, active series, retention, users, support, and export or migration paths.34

Infrastructure monitoring routes

Choose the route that matches operational ownership; Prometheus is open source, not free operations.

Infrastructure and application telemetry need one commercial suite

Compare Datadog and New Relic.

Both can connect infrastructure entities and telemetry to broader application monitoring workflows.

Verify: Agent coverage, entity model, telemetry meters, integration depth, and total cost require representative validation.13

Grafana and open collection are strategic but backends should be managed

Evaluate Grafana Cloud.

Managed storage and Grafana workflows can reduce backend operation while preserving an open ecosystem orientation.

Verify: Collectors, labels, dashboards, alerts, and usage governance remain operating work.2

The team deliberately owns the metrics system

Evaluate Prometheus.

Prometheus provides a self-operated pull-based metrics and alerting foundation with a broad ecosystem.

Verify: Capacity, retention, high availability, long-term storage, upgrades, security, and on-call ownership are not delegated.4

Boundary: This Task owns infrastructure resource and system metrics. It does not choose infrastructure, diagnose application request traces, group exceptions, run external checks, or define security audit evidence.

What actually differs

The decisive boundary is who operates collection and storage, then how the metrics model and suite scope affect cost and exit.

Ownership
Managed suites delegate more backend work; Prometheus transfers more reliability and capacity work to the team.24
Collection
Agents, integrations, exporters, and scrape models differ across resources and environments.4
Cardinality
Labels, custom metrics, active series, and retention can determine both usefulness and cost.12
Suite scope
Cross-signal correlation can help diagnosis while increasing vendor and billing coupling.12

Official resources

Verify current capabilities, collection boundaries, privacy controls, retention, and commercial terms in the official documentation before implementation.

DocumentationPrometheus overview

Sources

Official product documentation supporting the decision routes and boundaries on this page.

  1. 1
    Datadog Infrastructure Monitoring

    Datadog · Accessed Official

  2. 2
    Grafana Cloud infrastructure monitoring

    Grafana Labs · Accessed Official

  3. 3
    New Relic Infrastructure Monitoring

    New Relic · Accessed Official

  4. 4
    Prometheus overview

    Prometheus · Accessed Official