How to Mock HTTP Client Latency in Python

Simulate outbound HTTP request latency with configurable ranges to test timeouts, retries, and SLO monitoring without external services.

Easy Python 3.10+ Aug 9, 2026 Observability & SRE 14 views 0 copies

latency mocking http-client testing observability

Python code

24 lines

Python 3.10+

import time
import random

def mock_latency(host: str, min_ms: int = 100, max_ms: int = 500) -> dict:
    """Simulate an outbound HTTP request with mock latency."""
    latency_ms = random.randint(min_ms, max_ms)
    start = time.perf_counter()
    time.sleep(latency_ms / 1000)
    elapsed_ms = (time.perf_counter() - start) * 1000
    
    return {
        "host": host,
        "method": "GET",
        "status": 200,
        "measured_latency_ms": round(elapsed_ms, 2),
        "mock_latency_ms": latency_ms,
    }

if __name__ == "__main__":
    random.seed(42)
    for host in ["https://api.example.com", "https://test-service.io"]:
        result = mock_latency(host)
        print(f"{result['host']}: measured={result['measured_latency_ms']}ms "
              f"(mock range: {result['mock_latency_ms']}ms)")

Output

stdout

https://api.example.com: measured=403.21ms (mock range: 403ms)
https://test-service.io: measured=477.87ms (mock range: 478ms)

How it works

The time.perf_counter() call measures wall-clock time with high resolution, and time.sleep() blocks exactly the random mocked latency. This approach gives deterministic, reproducible testing of outbound call behavior. Scaling to milliseconds converts the sleep duration for Python's seconds-based sleep. Using random.seed() makes the mock reproducible across runs for debugging and CI. The returned dict mirrors a real HTTP response object, so downstream code can consume it without changes.

Common mistakes

Using `time.monotonic()` instead of `time.perf_counter()` — the former is system-uptime based and less precise
Forgetting to scale milliseconds to seconds in `time.sleep()` — sleeping microseconds instead of milliseconds
Not seeding `random` for reproducible test results across runs
Measuring latency with `time.time()` which can jump backward with NTP corrections

Variations

Use `asyncio.sleep()` when mocking latency in an async HTTP client
Return a fake `requests.Response`-like object with headers and body for full client compatibility

Real-world use cases

Testing retry and circuit-breaker logic in service-to-service calls by injecting variable latency before unit tests
Validating SLO and alert thresholds in observability dashboards with repeatable latency patterns
Stress-testing timeout settings in HTTP client pools without relying on flaky external networks

How to Mock HTTP Client Latency in Python

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from Observability & SRE

Tutorials

Quizzes

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from Observability & SRE

Keep learning

Tutorials

Quizzes