#📝 Note: Circuit Breaker Pattern

#Overview

The Circuit Breaker pattern prevents a system from repeatedly trying to execute an operation that is likely to fail, allowing it to recover gracefully without cascading failures across services.

#Core Idea

Inspired by electrical circuit breakers — when a downstream service is failing, "open the circuit" to stop sending requests, give it time to recover, and serve degraded/fallback responses instead.

#States

┌──────────┐     Failure threshold    ┌──────────┐
│  CLOSED  │ ──────exceeded────────▶ │   OPEN   │
│ (normal) │                          │ (failing)│
└──────────┘                          └─────┬────┘
     ▲                                      │
     │            ┌───────────┐             │
     │ Success    │ HALF-OPEN │  Timeout    │
     └────────────│ (testing) │◀────────────┘
                  └───────────┘
                     │ Failure
                     └──────────▶ Back to OPEN

• CLOSED: Requests flow normally. Failures are counted. If failures exceed threshold → switch to OPEN.
• OPEN: All requests immediately return fallback/error without calling downstream. After a timeout → switch to HALF-OPEN.
• HALF-OPEN: Allow a limited number of test requests through. If they succeed → CLOSED. If they fail → back to OPEN.

#Configuration Parameters

Parameter	Typical Value	Purpose
Failure Threshold	5 failures in 30s	When to trip the circuit
Open Timeout	30–60 seconds	How long to wait before testing recovery
Half-Open Max Requests	3–5 requests	How many test requests to allow
Success Threshold	3 consecutive	How many successes needed to close circuit

#When to Use

• Microservice Communication: Service A calls Service B; B goes down → A should degrade gracefully, not crash.
• External API Calls: Third-party APIs with unpredictable availability.
• Database Connections: Prevent connection pool exhaustion when DB is overloaded.
• ML Model Serving: If model server is slow/down, fallback to simpler logic (e.g., rules-only scoring).

#Fallback Strategies

• Cache: Serve stale cached data.
• Default Values: Return safe defaults (e.g., empty recommendations, approve low-risk transactions).
• Degraded Mode: Skip the failing component and proceed with partial functionality.
• Queue: Buffer requests for retry when the service recovers.

#Trade-offs

• Pros: Prevents cascade failures, faster failure detection, allows recovery time, improves overall system resilience.
• Cons: Adds complexity, fallback logic must be maintained, may mask underlying issues if not monitored.

#Used In This Repo

• E-Commerce (#11) — Search → Elasticsearch, Catalog → DB circuit breakers
• Recommendation Engine (#23) — Rec Service → Ranking Service fallback to cached/popular recs
• Fraud Detection (#24) — Scoring Service → ML Scorer fallback to rules-only scoring