Generate Synthetic CPU Utilization Metrics in Python

Creates realistic time-series CPU utilization samples with timestamps, noise, and output as structured JSON for observability demos and testing.

Easy Python 3.9+ Aug 9, 2026 Observability & SRE 14 views 0 copies

observability metrics time-series synthetic-data json datetime

Python code

37 lines

Python 3.9+

from datetime import datetime, timedelta
import random
import json


def generate_metric_samples(base_value, noise, count=60, interval_minutes=1):
    """Generate realistic CPU utilization samples for a given time window."""
    timestamps = []
    values = []

    now = datetime.utcnow()
    start_time = now - timedelta(minutes=(count - 1) * interval_minutes)

    for i in range(count):
        timestamp = start_time + timedelta(minutes=i * interval_minutes)
        value = base_value + random.uniform(-noise, noise)
        timestamps.append(timestamp.isoformat())
        values.append(round(max(0, min(100, value)), 2))

    return timestamps, values


if __name__ == "__main__":
    # Help beginners understand the shape of time-series data
    ts, vals = generate_metric_samples(base_value=65.0, noise=10.0, count=5, interval_minutes=1)

    dataset = {
        "metric": "cpu_utilization",
        "unit": "percent",
        "host": "web-server-01",
        "samples": [
            {"timestamp": ts[i], "value": vals[i]}
            for i in range(len(ts))
        ],
    }

    print(json.dumps(dataset, indent=2))

Output

stdout

{
  "metric": "cpu_utilization",
  "unit": "percent",
  "host": "web-server-01",
  "samples": [
    {
      "timestamp": "2025-03-23T12:34:56.789012",
      "value": 68.42
    },
    {
      "timestamp": "2025-03-23T12:35:56.789012",
      "value": 63.51
    },
    {
      "timestamp": "2025-03-23T12:36:56.789012",
      "value": 70.89
    },
    {
      "timestamp": "2025-03-23T12:37:56.789012",
      "value": 59.37
    },
    {
      "timestamp": "2025-03-23T12:38:56.789012",
      "value": 66.73
    }
  ]
}

How it works

The function generate_metric_samples computes a start time by subtracting count - 1 intervals from the current UTC time, then iterates to create timestamps at fixed intervals. Each value is the base plus a uniform random noise between -noise and +noise, then clipped to 0–100 and rounded to two decimals. The __main__ block builds a dictionary with metadata (metric name, unit, host) and a list of sample dicts, then serializes to JSON with json.dumps(indent=2) for readability. This mimics the structure of many monitoring APIs and time-series databases like Prometheus or InfluxDB.

Common mistakes

Using `datetime.now()` instead of `datetime.utcnow()` leading to timezone confusion in logs
Not clipping values to a realistic range (0–100 for CPU), producing nonsensical metrics
Forgetting to round values, resulting in noisy floating-point artifacts
Hardcoding count or interval instead of making them parameters

Variations

Use `datetime.now(timezone.utc)` for timezone-aware timestamps instead of naive UTC
Generate samples with a seasonal pattern using a sine wave plus noise

Real-world use cases

Creating synthetic CPU metrics to test dashboards in Grafana or Datadog during development.
Feeding realistic time-series data into alerting systems to validate threshold rules.
Supporting load tests by generating expected metric patterns for capacity planning.

Generate Synthetic CPU Utilization Metrics 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