Design and Operate a Monolith on Kubernetes

Context

You are joining an infrastructure team. The backend is a single monolithic service. The company is not pursuing microservices or aggressive horizontal scaling right now. Assume:

• Moderate scale (e.g., peak 2–5k RPS), high availability target across multiple zones, zero-downtime deploys.
• Single cloud provider with managed Kubernetes, managed databases/caches/object storage where possible.
• Mix of web (HTTP), background workers, and scheduled jobs from the same monolith codebase.

Task

Design how you would run this monolith on Kubernetes. Cover:

1. Deployment topology

• Pods, replicas, separation of web vs worker vs cron jobs
• Node pools, placement, disruption policies

2. Networking and traffic management

• Services, Ingress, TLS, rate limiting, canary traffic splitting

3. Configuration and secret management

• Configs, secrets, rotation, feature flags

4. CI/CD and rollout strategies

• Pipelines, blue/green vs canary, automated rollback

5. Resource management and autoscaling

• Requests/limits, HPA/VPA, cluster autoscaling, capacity buffers

6. Handling stateful dependencies

• Databases, caches, object storage, migrations, connections

7. Observability

• Logs, metrics, traces, SLOs, alerting

8. Incident response

• Runbooks, on-call, kill switches, rollback

9. Disaster recovery and backups

• RTO/RPO targets, backups, restore drills

10. Architecture trade-offs

• Monolith vs microservices, concrete signals that justify a change, and a low-risk, incremental migration path (service boundaries, data ownership, API contracts, team workflows)