HARD

Design a Metrics Collection System

Design a metrics collection and monitoring system like Datadog or Prometheus handling millions of time-series data points.

Estimated Time: 45 minutes
#Time Series#Aggregation#Alerting#Visualization
Solution Overview

Use time-series database optimized for write-heavy workloads. Implement pre-aggregation for common queries. Support flexible alerting rules.

Hints to Get Started
1

Time-series compression techniques

2

Cardinality explosion handling

3

Push vs pull model

Storage
  • Hot storage (recent)
  • Warm storage (aggregated)
  • Cold storage (raw archive)
Data Model

metric_name{labels} value timestamp

Query Patterns
  • Range queries
  • Aggregations
  • Top-K