How to Compute SRE Metrics Like Error Rate and Availability in Python

from collections import deque
from datetime import datetime, timedelta
from typing import Dict, Deque


class LogMetrics:
    """Simple observability helper to track log events and calculate SRE metrics."""

    def __init__(self, window_seconds: int = 60):
        self.window_seconds = window_seconds
        self.events: Deque[tuple] = deque()
        self.counts: Dict[str, int] = {}

    def add_event(self, level: str) -> None:
        """Record a log event at the current timestamp."""
        now = datetime.now()
        self.events.append((now, level))
        self.counts[level] = self.counts.get(level, 0) + 1
        self._prune_expired(now)

    def _prune_expired(self, current_time: datetime) -> None:
        """Remove events older than the sliding window."""
        cutoff = current_time - timedelta(seconds=self.window_seconds)
        while self.events and self.events[0][0] < cutoff:
            _, level = self.events.popleft()
            self.counts[level] = max(0, self.counts[level] - 1)
            if self.counts[level] == 0:
                del self.counts[level]

    def error_rate_per_second(self) -> float:
        """Calculate error rate (errors per second) in the current window."""
        if not self.events:
            return 0.0
        total_errors = self.counts.get("ERROR", 0)
        elapsed = (self.events[-1][0] - self.events[0][0]).total_seconds()
        return total_errors / elapsed if elapsed > 0 else 0.0

    def availability(self) -> float:
        """Calculate availability percentage (non-error events / total events)."""
        total = max(1, sum(self.counts.values()))
        errors = self.counts.get("ERROR", 0)
        return (total - errors) / total * 100


if __name__ == "__main__":
    metrics = LogMetrics(window_seconds=10)

    # Simulate a series of events: mostly INFO, with a few ERRORs
    for _ in range(25):
        metrics.add_event("INFO")
    for _ in range(5):
        metrics.add_event("ERROR")

    print(f"Events in window: {sum(metrics.counts.values())}")
    print(f"Error rate (errors/sec): {metrics.error_rate_per_second():.2f}")
    print(f"Availability: {metrics.availability():.1f}%")