BACK TO SELECTED WORK
CASE STUDY 03 / 09Engineering / Blueprint
Distributed performance testing

Indus Tester

A Go-native distributed load-testing platform for CI/CD pipelines, staging reliability checks, load balancer validation, and performance gates.

GogRPCPrometheusHDR HistogramsCI/CDJSON Reports
Indus Tester project previewPROJECT / BLUEPRINT
01 / THE WORK

Behind the build.

Indus Tester models load over time instead of simple request loops. A controller coordinates multiple agents, distributes virtual users, streams metrics, evaluates thresholds, and produces human-readable plus machine-readable reports. It is designed for real engineering workflows where p95 latency, throughput, error ratio, and CI pass/fail behavior matter.

02 / ENGINEERING DECISIONS

What makes it work.

01 / CAPABILITYTime-driven load profiles including constant, ramp, step, and spike
02 / CAPABILITYController-agent architecture for distributed test execution
03 / CAPABILITYp50, p90, p95, p99, RPS, error ratio, and time-series metrics
04 / CAPABILITYThreshold-based exit codes for CI/CD performance gates
03 / INTERACTIVE WALKTHROUGH

From input to outcome.

STAGE 1 OF 5

Define scenarios, target endpoints, stages, load, and thresholds.

04 / ARCHITECTURE

Under the hood.

Distributed Controller Node & Worker Agent Topology

Topology Diagram 3.1
Control Plane

Controller Master

Distributes test scenarios, coordinates virtual user schedules, and evaluates threshold gates.

Data Plane

gRPC Worker Agents

Executes constant, ramp, step, or spike load profiles against target load balancers and endpoints.

Analysis Engine

HDR Histograms & Reports

Captures p50/p90/p95/p99 latency distributions and outputs JSON reports for CI/CD gates.

05 / RESULTS & USE CASES

Built to matter.

OUTCOME 01Repeatable performance validation before release
OUTCOME 02Fast feedback inside CI pipelines
OUTCOME 03Clear reports for debugging regressions
Backend teamsDevOps pipelinesStaging validationLoad balancer testing