Vegeta — HTTP Load Testing Tool at Constant Request Rates

What it is

Vegeta is a Go-based HTTP load testing tool that attacks endpoints at a precise constant request rate. Unlike tools that ramp up connections until the server breaks, Vegeta maintains the rate you specify and measures how the server responds. It produces HDR histograms, JSON reports, latency distributions, and visual plots.

Vegeta targets backend engineers, SREs, and performance testers who need reproducible load tests with accurate latency measurements.

How it saves time or tokens

Vegeta's constant-rate approach produces more meaningful results than burst-based tools. You know exactly how many requests per second were sent, making it easy to compare runs and identify regressions. The CLI piping model (attack | encode | report) is composable and scriptable.

Vegeta is a single binary with zero dependencies. No JVM, no config files, no cluster setup.

How to use

1. Install Vegeta: go install github.com/tsenart/vegeta@latest or brew install vegeta
2. Create a targets file with HTTP endpoints
3. Run an attack at a specified rate
4. Generate reports from the results

Example

# Attack at 100 requests/second for 30 seconds
echo 'GET https://api.example.com/health' | \
  vegeta attack -rate=100 -duration=30s | \
  vegeta report

# Output:
# Requests      [total, rate, throughput]  3000, 100.03, 99.97
# Duration      [total, attack, wait]      30.01s, 29.99s, 12.1ms
# Latencies     [min, mean, 50, 90, 95, 99, max]  2.1ms, 12.3ms, 10.1ms, 22.4ms, 35.6ms, 89.2ms, 210.1ms
# Success       [ratio]                   99.87%
# Status Codes  [code:count]              200:2996  500:4

# Generate a latency plot
echo 'GET https://api.example.com/health' | \
  vegeta attack -rate=100 -duration=30s | \
  vegeta plot > latency.html

# Multiple endpoints with headers
cat <<EOF | vegeta attack -rate=50 -duration=10s | vegeta report
GET https://api.example.com/users
Authorization: Bearer token123

POST https://api.example.com/orders
Content-Type: application/json
@body.json
EOF

Related on TokRepo

• Testing tools -- Load testing and quality assurance
• DevOps tools -- Infrastructure and performance tools

Common pitfalls

• Running Vegeta from a machine with limited bandwidth distorts results; test from a host near the target
• Very high rates (10K+/s) may exhaust local ports; increase ulimit -n and use connection keep-alive
• Vegeta measures client-side latency including network time; compare results from the same network location