Begum

Begum is a tool that generates metrics similar to what would come out of an HTTP-based web or microservice. These metrics can be manipulated by making API calls to add and remove instances, adjust latencies, and error rates.

By making various API calls you can simulate a cluster of instances of a service that generate realistic metrics, then cause some of these to become stricken with errors or increased latency.

Instances

Begum runs as many "instances" as you like. Each has a name — by default sequentially A, B, C… etc and emits metrics.

JSON

{
  "error_pct": 1,
  "latency_min_ms": 100,
  "latency_max_ms": 300,
  "latency_offset_ms": 0
}

Fields And Effects

error_pct: % of requests that should be errors
latency_min_ms: Minimum value of latency per request
latency_max_ms: Maximum value of latency per request
latency_offset_ms: Amount of extra latency added after choosing a random latency between latency_min_ms and latency_max_ms.

API

DELETE /instance/X removes an instance named X, no body.
POST /instance/X adds an instance named X, expects JSON
PUT /instance/X adjusts the parameters of an instance named X, expects JSON

Metrics Emitted

The following metrics are emitted:

request_duration_millis with tags/labels code and instance with quantiles 0.5, 0.9, 0.99. Also includes request_duration_millis_count and request_duration_millis_sum.
errors_encountered_total with tags/labels code and instance.

Name

Begum is a female royal and aristocratic title from Central and South Asia.