Cruxly › Cloud and DevOps interview preparation

Cloud and DevOps interview preparation

DevOps and cloud interviews look for judgement under failure: how you ship safely, observe a system, and respond when it breaks. These lessons cover each area with concrete trade-offs.

Cloud & DevOps: Containers and Orchestration

What a container isolates, why a pod (not a container) is the real scheduling unit, and the failure modes that only show up under load.

• Resource Requests and Limits: Sizing a Pod That Survives Real TrafficOOMKills, CPU throttling, and the noisy-neighbor problem
• Liveness, Readiness, and a Rollout That Doesn't Take the Service Downthe probe misconfiguration that causes its own outage

Cloud & DevOps: CI/CD Pipelines

Build-once/promote-everywhere, real rollback mechanics, and the CI pipeline's own supply-chain attack surface.

• Build Once, Promote Everywhere: Designing a Pipeline Without Environment Driftimmutable artifacts, digest pinning, and the rebuild trap
• Deployment Strategies and the Rollback Plan That Survives a Real Incidentblue-green, canary, and what 'roll back' actually leaves behind
• Pipeline Security: The CI System as Its Own Attack Surfacedependency pinning, SBOMs, and why a compromised runner beats a compromised prod box

Cloud & DevOps: Infrastructure as Code

Terraform state as the file that can take down your infrastructure, real locking mechanics, and the drift a manual console change leaves behind.

• Terraform State: The File That Can Take Down Your Infrastructurelocking, corruption, and why state is not just a cache
• Idempotency, Plan Review Discipline, and the Drift a Console Change Leaves Behindwhy terraform apply twice must be safe, and the incident-week landmine

Cloud & DevOps: Observability

Metrics vs logs vs traces as different questions with different costs, cardinality explosions, and alerting on user-experienced symptoms instead of internal causes.

• Metrics, Logs, and Traces: Picking the Pillar That Actually Answers Your Questiondifferent questions, different costs, and the cardinality trap
• SLOs, Error Budgets, and Alerting on What Users Actually Experiencethe Google SRE discipline for deciding what's worth a page

Cloud & DevOps: Networking

DNS TTL trade-offs, L4 vs L7 load balancing, retry storms, and what a service mesh's sidecar tax actually costs.

• DNS Failover, Load Balancer Layers, and the Retry Storm That Follows an OutageTTL trade-offs, L4 vs L7, and thundering-herd retries
• Service Mesh: What a Sidecar Proxy Actually Buys You, and What It CostsmTLS, uniform telemetry, and the sidecar tax

Cloud & DevOps: Cloud Cost Engineering

Spot's real discount versus its advertised one, cross-AZ/egress charges that never show up in a latency dashboard, and cost attribution that survives a shared bill.

• Reserved, Spot, and On-Demand: The Real Trade-off Behind Each Discountadvertised savings vs realistic savings, and commitment risk
• The Bill You Don't See Coming: Egress, Cross-AZ Charges, and Real Cost Attributiontraffic that costs money without ever looking slow

Cloud & DevOps: Security

Least-privilege IAM and the real blast radius of getting it wrong, and why environment variables are a genuine rotation anti-pattern, not just a convenience shortcut.

• Least-Privilege IAM and the Blast Radius of Getting It Wrongthe Capital One pattern, and why broad roles are the default outcome
• Secrets Management: Why Environment Variables Aren't Enoughauditability, rotation friction, and dynamic secrets

Cloud & DevOps: Incident Response

The Incident Commander role and why it isn't the resolver, and blameless postmortems that produce action items with owners instead of a document nobody revisits.

• Running an Incident: Roles, Communication, and Not Making It Worsewhy the Incident Commander is deliberately not the resolver
• Blameless Postmortems That Actually Produce Working Action Itemsthe Etsy/Dekker origin, and why most postmortems don't change anything