Grafana — Open Source Data Visualization & Observability

What Grafana Does

• Dashboards: Build interactive dashboards with 50+ panel types (graphs, gauges, tables, heatmaps, maps)
• Data Sources: Connect 100+ data sources including Prometheus, InfluxDB, Elasticsearch, PostgreSQL, MySQL, Loki, Tempo, and cloud services
• Alerting: Define alert rules, notification channels (Slack, PagerDuty, email), and escalation policies
• Explore: Ad-hoc query interface for investigating metrics and logs in real-time
• Annotations: Mark events on graphs (deploys, incidents, changes) for correlation
• Variables: Template variables for dynamic, reusable dashboards
• Provisioning: Infrastructure-as-code dashboard and datasource management via YAML/JSON

Architecture

┌──────────────┐     ┌──────────────┐     ┌──────────────┐
│  Browser     │────▶│  Grafana     │────▶│  Data Sources│
│  Dashboard   │     │  Server (Go) │     │  Prometheus  │
└──────────────┘     └──────────────┘     │  InfluxDB    │
                                          │  Elasticsearch│
                                          │  PostgreSQL  │
                                          │  Loki (Logs) │
                                          │  Tempo (Traces)│
                                          └──────────────┘

Self-Hosting

Docker Compose with Prometheus

services:
  grafana:
    image: grafana/grafana-oss:latest
    ports:
      - "3000:3000"
    environment:
      GF_SECURITY_ADMIN_PASSWORD: your-password
    volumes:
      - grafana-data:/var/lib/grafana
    depends_on:
      - prometheus

  prometheus:
    image: prom/prometheus:latest
    ports:
      - "9090:9090"
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml
      - prometheus-data:/prometheus

volumes:
  grafana-data:
  prometheus-data:

Key Features

Dashboard Panels

Available panel types:
├── Time Series — line/area/bar charts over time
├── Stat — single value with sparkline
├── Gauge — circular gauge with thresholds
├── Bar Chart — categorical comparisons
├── Table — tabular data with sorting/filtering
├── Heatmap — time-based density visualization
├── Geomap — geographic data on world map
├── Logs — log line viewer with search
├── Traces — distributed trace visualization
├── Node Graph — network topology
├── Canvas — custom layout with drag-and-drop
└── 40+ more community panels

PromQL Example (Prometheus)

# CPU usage per container
rate(container_cpu_usage_seconds_total{namespace="production"}[5m]) * 100

# Request rate
sum(rate(http_requests_total[5m])) by (status_code)

# P99 latency
histogram_quantile(0.99, rate(http_request_duration_seconds_bucket[5m]))

# Error rate percentage
sum(rate(http_requests_total{status=~"5.."}[5m])) / sum(rate(http_requests_total[5m])) * 100

Alerting

# Alert rule example
groups:
  - name: critical
    rules:
      - alert: HighErrorRate
        expr: sum(rate(http_requests_total{status=~"5.."}[5m])) / sum(rate(http_requests_total[5m])) > 0.05
        for: 5m
        labels:
          severity: critical
        annotations:
          summary: "Error rate above 5%"

Dashboard as Code

{
  "dashboard": {
    "title": "My Dashboard",
    "panels": [
      {
        "title": "CPU Usage",
        "type": "timeseries",
        "datasource": "Prometheus",
        "targets": [
          {
            "expr": "rate(node_cpu_seconds_total{mode='idle'}[5m])",
            "legendFormat": "{{instance}}"
          }
        ]
      }
    ]
  }
}

LGTM Stack (Full Observability)

Grafana Labs provides a complete open-source observability stack:

Grafana vs Alternatives

常见问题

Q: Grafana 适合初学者吗？ A: Grafana 本身的学习曲线不高——拖拽面板、选择数据源即可。难点在于理解数据源（如 Prometheus 的 PromQL 查询语言）。建议从导入社区 Dashboard 开始学习。

Q: Grafana OSS 和 Grafana Cloud 有什么区别？ A: OSS 是完全免费的自托管版本。Cloud 提供托管服务，免费层包含 10K 指标/50GB 日志/50GB 追踪。Cloud 还包含一些 OSS 没有的功能如 ML 驱动告警。

Q: 可以嵌入到自己的应用中吗？ A: 可以。Grafana 支持 iframe 嵌入和匿名访问配置。也可以使用 Grafana 的 API 获取面板截图。

来源与致谢

• GitHub: grafana/grafana — 73.1K+ ⭐ | AGPL-3.0
• 官网: grafana.com