Trace

Abstract

---

• Traces help us track the flow of requests through various services and components of the system
• A trace is made up one or more Span
• Once we identify the span for latency, we can proceed with optimisation
• We can use tools like Datadog APM, Tempo and Zipkin etc

Terminologies

---

Runtime Metrics

• Allow to view Heap Segment, non-heap memory usage and Garbage Collector activity of the app
• Datadog can have this enabled with export DD_RUNTIME_METRICS_ENABLED=true

Instrumented

• Code or tools have been added to the application to monitor, measure, or analyze its behavior during execution
• Provide insights into the application’s performance, functionality, and other operational characteristics
• This is particularly useful for debugging, performance tuning, and monitoring purposes

Pipeline

• Datadog example

Host-side

• We can tune the Sampling
  • Library Sampling overrides Agent Sampling
• Trace Metrics are the Metrics, directly connected Instrumented application, calculated based on 100% of the app’s traffic

Datadog backend side

• Live Search allows us to search Span using any tag or Span
• Generate Custom Metric from Span
• Retention Filters - how long we want to retain the trace
• Dashboard used to give a visual representation of the app for optimisation and debugging