Перейти до основного вмісту

observability-stack

ADR 009: Prometheus & Grafana for Metrics, ELK for Logs, Zipkin for Tracing​

Context:
We need full‑stack observability: real‑time metrics, log aggregation, and distributed tracing.

Decision:

  • Metrics: Prometheus + Grafana dashboards
  • Logs: ELK Stack (Filebeat → Elasticsearch → Kibana)
  • Tracing: Zipkin via Spring Cloud Sleuth

Consequences:

  • Pros:
    • Unified visibility: metrics, logs, traces correlated by trace IDs.
    • Highly customizable alerting and dashboarding.
  • Cons:
    • Operational overhead: multiple clusters, storage sizing, retention policies.
    • Potential performance impact of sampling/tracing on high‑throughput flows.