Circuit Breaker

Protects the system from cascading failures by cutting off calls to a struggling service.

Context

A dependency (external API, DB, cache) starts to fail/timeout. Your application keeps calling it, exhausting threads and propagating latency.

Problem

Calling something known to be broken wastes resources.
Massive retries worsen the incident.

Solution

A breaker monitors failures. If they exceed a threshold, it transitions to Open and immediately rejects calls. After a period, it attempts in Half-Open and decides.

#Example in C# — Polly

using Polly;
using Polly.CircuitBreaker;

var breaker = Policy
    .Handle<HttpRequestException>()
    .CircuitBreakerAsync(
        exceptionsAllowedBeforeBreaking: 5,
        durationOfBreak: TimeSpan.FromSeconds(30),
        onBreak: (ex, ts) => log.LogWarning("Breaker opened {Span}", ts),
        onReset: () => log.LogInformation("Breaker closed"));

var response = await breaker.ExecuteAsync(() => httpClient.GetAsync("/api/orders"));

Tradeoffs

Pro	Con
Prevents cascading failures	Some requests fail fast even if the service has recovered
Allows remote service time to recover	Configuring thresholds requires observability

#Variants

Bulkhead (isolates thread pools per dependency).
Retry with exponential jitter (often combined with Circuit Breaker).

#extra #resilience