monitoring-expert

Monitoring involves collecting metrics, logs, and traces from your services, then visualizing them with tools like Prometheus and Grafana to detect issues early. Set up Prometheus scraping, define meaningful alerting rules, and configure Grafana dashboards to track RED metrics (request rate, errors, duration) so you can respond to incidents rapidly.

Metrics are numerical measurements aggregated over time (counters, gauges, histograms); logs are discrete events with context; traces follow requests across services. Together they form observability: metrics show *what* happened at scale, logs show *why*, and traces show *how* requests flow. Use all three for complete visibility.

Instrument your service by adding Prometheus client libraries to export metrics and OpenTelemetry SDKs to capture traces. Configure collectors to scrape endpoints, define meaningful metric types (counter for totals, gauge for current state, histogram for latencies), and export traces to your backend. Visualize everything in Grafana dashboards.

Yes. Prometheus stores time-series metrics and Grafana visualizes them. Build RED/USE dashboards in Grafana querying Prometheus to analyze request latency, error rates, and resource utilization. This combination enables rapid performance optimization and incident correlation.

Correlation IDs are unique identifiers attached to requests across all services and logs. They link related log entries and traces together, enabling you to reconstruct request flows and debug issues faster. Pair them with structured JSON logging to make log parsing and alerting reliable.

Use load testing tools like k6, Artillery, or Locust to simulate traffic and measure performance under stress. Monitor metrics and traces during tests to identify bottlenecks, then use profiling data to optimize code. This feeds into capacity planning and helps set realistic alerting thresholds.