SysPadExamples › Metrics monitoring & alerting

Metrics monitoring & alerting

Agents push metrics through a collection stream into a time-series DB; a Query Service feeds dashboards while an Alert Manager evaluates rules against the same store and pages out.

One TSDB serving both a dashboard read path and an independent alert-evaluation path.

Metric Agents Users Collection Stream Kinesis Aggregator Lambda TSDB Timestream Query Service Lambda Dashboards QuickSight Alert Manager Lambda Alert Topic SNS AT LIMIT PagerDuty External

How it works

Metric Agents
Every host and service runs a collection agent that pushes metrics on a fixed interval - this is a push pipeline, so the agents are the front door, not a client behind a load balancer.
Collection Stream
A durable buffer between a huge fan-in of agents and the aggregator, so a burst (a deploy, a fleet-wide GC pause) is absorbed as stream backlog instead of dropped datapoints.
TSDB
The superposition centre: the ingest flow writes it, the dashboard flow scans it for charts, and the alert flow scans it to evaluate rules - three journeys stacking load on one store.
Alert Manager
Re-evaluates every alert rule against the same TSDB on a short cycle, independent of the dashboard path - most evaluations find nothing, so only a small fraction go on to page out.

Request flows

Collect metrics

  1. Metric Agents
  2. Collection Stream
  3. Aggregator
  4. TSDB

View a dashboard

  1. TSDB
  2. Query Service
  3. Dashboards

Evaluate alert rules

  1. TSDB
  2. Alert Manager

Fire an alert

  1. Alert Manager
  2. Alert Topic
  3. PagerDuty

Other interview practice

Open this architecture in SysPad · All examples