Jonathan Lalou's Blog

1) Patterns for Graceful Degradation

When dependencies fail, your system should still provide partial service. Examples:

• Show cached product data if the pricing service is down.
• Allow “read-only” mode if writes are failing.
• Provide degraded image quality if the CDN is unavailable.

2) Circuit Breakers

Prevent cascading failures with Resilience4j or Hystrix:

@CircuitBreaker(name = "inventoryService", fallbackMethod = "fallbackInventory")
public Inventory getInventory(String productId) {
    return restTemplate.getForObject("/inventory/" + productId, Inventory.class);
}

public Inventory fallbackInventory(String productId, Throwable t) {
    return new Inventory(productId, 0);
}

3) Retries with Backoff

Retries should be bounded and spaced out:

@Retry(name = "paymentService", fallbackMethod = "fallbackPayment")
public PaymentResponse processPayment(PaymentRequest req) {
    return restTemplate.postForObject("/pay", req, PaymentResponse.class);
}

RetryConfig config = RetryConfig.custom()
    .maxAttempts(3)
    .waitDuration(Duration.ofMillis(200))
    .intervalFunction(IntervalFunction.ofExponentialBackoff(200, 2.0, 0.5)) // jitter
    .build();

4) Scaling Microservices in Kubernetes/ECS

Scaling is not just replicas—it’s smart policies:

• Kubernetes HPA: Scale pods based on CPU or custom metrics (e.g., p95 latency).

  kubectl autoscale deployment api --cpu-percent=70 --min=3 --max=10

• ECS: Use Service Auto Scaling with CloudWatch alarms on queue depth.
• Pre-warm caches: Scale up before big events (e.g., Black Friday).