Monitoring

Why monitoring matters

Reliability starts here. Immediate signals (alerts), trends (capacity planning), UX 개선.
장애를 줄이고, 고객 문제를 사전에 발견. (reduce MTTR / prevent incidents)

Cloud Monitoring architecture

Layer	What it does	Google Cloud bits
Collection	Collects metrics/logs/traces	GKE, Compute Engine, App Engine, Cloud Ops Agent, GMP, OTEL
Storage & routing	Stores + routes to viz/analysis	Cloud Monitoring API (metric ingest & routing)
Analysis & Viz	Dashboards, alerts, uptime	Dashboards, Metrics Explorer, Uptime checks, Alerting policies, Notifications

Platform monitoring

기본 제공 시스템 메트릭: 1500+ metrics / 100+ services, VM per-instance 25+ metrics 수준.
무료 시스템 메트릭 for GCP services. (과금: 커스텀/수집 방식에 따라 달라짐)
이미 다른 3rd-party(예: Datadog/New Relic) 쓰면 API 통합으로 GCP 메트릭 가져가면 됨.

Application monitoring - GKE (Prometheus way)

Many users want Prometheus. Use Google Managed Prometheus (GMP).
- Prometheus exposition ingest, PromQL, expression browser, rule eval compatible.
- GKE cluster + workload metrics exposed as Prometheus data, but lives in Cloud Monitoring.
추천: GMP for GKE workloads.

Application monitoring - Compute Engine (VMs)

Use Cloud Ops Agent to collect in-process + 3rd-party app metrics on VMs.
- 30+ plugins, richer OS metrics (Windows/Linux 다양한 배포판).
- OpenTelemetry 기반 → custom apps = OTEL SDK로 계측 → custom metrics 수집 가능.
If broader ecosystem fit is needed: Datadog/New Relic 가능 (GCP 시스템 메트릭은 native API로 수집).

Hybrid monitoring & logging

Partner BindPlane (Blue Medora): on-prem + AWS/Azure/Alibaba/IBM Cloud → single pane in Cloud Monitoring.
- 150+ data sources integration.
- 추가 라이선스 비용 없음 (but: BindPlane metrics = custom metrics → 과금, logs = Cloud Logging 요율 적용).

Monitoring multiple projects

기본 동작 (Default metrics scope = current project)