Observability MOC

Properties

tags cs cs/swe cs/swe/observability

created 14.04.2026, 17:00

modified 02.06.2026, 09:21

published Empty

sources OpenTelemetry Docs · Grafana LGTM Stack

topics Observability

authors Empty

ai-assisted Yes

Ability to understand the internal state of a system byw examining its external outputs
Three pillars: logs (events), metrics (measurements), traces (request journeys)
Critical for operating distributed/microservice systems in production

# Three Pillars

Logs — discrete timestamped events; what happened and when
- Structured (JSON) logs are far more queryable than plain text
Metrics — numeric time-series data; aggregated measurements over time
- e.g. request rate, error rate, CPU %, p99 latency
Traces — end-to-end request path across services; the call graph with timings
- Answer: which service caused this 2s latency?

# The LGTM + OTel Architecture

1
2
3
4
5
6
7
8
Your Services
  │ (OTel SDK — auto or manual instrumentation)
  ▼
Grafana Alloy / OTel Collector
  │
  ├──► Loki       (logs)      ──┐
  ├──► Mimir      (metrics)   ──┤──► Grafana (unified UI)
  └──► Tempo      (traces)    ──┘

OpenTelemetry — instrumentation standard; vendor-neutral SDK + Collector
LGTM Stack — Grafana’s storage and visualization stack

# Sections

OpenTelemetry
- Vendor-neutral instrumentation standard; SDK, OTel Collector, OTLP protocol
LGTM Stack
- Loki + Grafana + Tempo + Mimir; deployment via Helm, Grafana Alloy
Distributed Tracing
- Spans, traces, context propagation across services

Observability MOC

# Three Pillars

# The LGTM + OTel Architecture

# Sections

Backlinks