Skip to main content

KPIs

Key Performance Indicators (KPIs) help you measure the health and reliability of your systems. Unified Monitoring provides the data to implement SLI/SLO‑style KPIs across services.

Core Golden Signals​

For most user‑facing services, start with the classic golden signals:

  • Latency – response time for key operations or APIs.
  • Traffic – request volume or throughput.
  • Errors – error rate or failure percentage.
  • Saturation – capacity usage such as CPU, memory, queue depth, or connection counts.

These can be derived from cloud‑native metrics (for example, ELB/ALB metrics, API Gateway metrics) or host‑level metrics via Breeze/Zabbix.

Additional KPIs​

Depending on your environment, consider:

  • Availability/uptime – percentage of time a service is meeting its availability target.
  • Error budget burn – rate at which you are consuming error budget.
  • Dependency health – status of databases, message queues, caches, and external APIs.
  • Capacity headroom – how close key resources are to capacity limits.
  • Alert noise – number of alerts per service or per on‑call shift.

Using KPIs in Unified Monitoring​

  • Define metrics and labels that correspond to each KPI.
  • Build dashboards that show current value, recent history, and trends.
  • Create alert policies that fire when KPIs breach agreed thresholds or burn rates.
  • Use CMDB attributes (application, environment, owner) to scope KPIs by service or business unit.