What is a key consideration when designing applications or infrastructure in cloud native practices?

What is a key consideration when designing applications or infrastructure in cloud native practices?

• A. Potential for component failure
• B. Always assuming zero failures
• C. Only consider performance, not reliability
• D. Ignore distributed tracing and resilience

Answer: (A)

In cloud-native design, systems are distributed and prone to partial failures, so the main idea is to design with failure in mind. You build for resilience by adding redundancy, keeping services stateless where possible, and enabling graceful degradation when a component falters. Use patterns like timeouts with backoff, retries, and circuit breakers to prevent a single failing part from cascading into others. Ensure operations are idempotent so retries don’t cause duplicates, and isolate faults with clear boundaries (bulkheads). Observability is essential, including distributed tracing, logs, and metrics, so you can detect and diagnose issues quickly. Kubernetes helps by restarting failed pods and routing traffic, but it doesn’t replace the need for robust resilience patterns and testing, such as chaos engineering, to verify reliability. The other options miss reliability: assuming zero failures, focusing only on performance, or ignoring tracing and resilience would leave you unprepared for real-world conditions.

In cloud-native design, systems are distributed and prone to partial failures, so the main idea is to design with failure in mind. You build for resilience by adding redundancy, keeping services stateless where possible, and enabling graceful degradation when a component falters. Use patterns like timeouts with backoff, retries, and circuit breakers to prevent a single failing part from cascading into others. Ensure operations are idempotent so retries don’t cause duplicates, and isolate faults with clear boundaries (bulkheads). Observability is essential, including distributed tracing, logs, and metrics, so you can detect and diagnose issues quickly. Kubernetes helps by restarting failed pods and routing traffic, but it doesn’t replace the need for robust resilience patterns and testing, such as chaos engineering, to verify reliability. The other options miss reliability: assuming zero failures, focusing only on performance, or ignoring tracing and resilience would leave you unprepared for real-world conditions.