Cloud Infrastructure Skill Skill

Cloud Infrastructure Skill

Quick Reference

| Platform | Market | Best For | Learning | |----------|--------|----------|----------| | AWS | 32% | Everything | 3-6 mo | | Azure | 24% | Microsoft stack | 3-6 mo | | GCP | 11% | Data, ML | 3-6 mo | | Cloudflare | Edge | CDN, Workers | 2-4 wk |

Learning Paths

AWS

[1] IAM + VPC (1-2 wk)
 │  └─ Roles, policies, networking
 │
 ▼
[2] Compute: EC2, Lambda (2-3 wk)
 │
 ▼
[3] Storage: S3, EBS (1-2 wk)
 │
 ▼
[4] Database: RDS, DynamoDB (2-3 wk)
 │
 ▼
[5] Containers: ECS, EKS (3-4 wk)
 │
 ▼
[6] Monitoring: CloudWatch (1-2 wk)

Docker & Containers

[1] Docker Basics (1 wk)
 │  └─ Images, containers, Dockerfile
 │
 ▼
[2] Multi-stage Builds (1 wk)
 │  └─ Optimization, layer caching
 │
 ▼
[3] Docker Compose (1 wk)
 │  └─ Multi-container apps
 │
 ▼
[4] Registry & Security (1 wk)
    └─ Push/pull, scanning, non-root

Kubernetes

[1] Pods & Deployments (2 wk)
 │
 ▼
[2] Services & Networking (1-2 wk)
 │
 ▼
[3] ConfigMaps & Secrets (1 wk)
 │
 ▼
[4] Helm Charts (2 wk)
 │
 ▼
[5] Production Patterns (ongoing)
    └─ HPA, PDB, resource limits

Terraform (IaC)

[1] Resources & State (1 wk)
 │
 ▼
[2] Variables & Outputs (1 wk)
 │
 ▼
[3] Modules (1-2 wk)
 │
 ▼
[4] Remote State (1 wk)
 │
 ▼
[5] Workspaces & Environments (1 wk)

Kubernetes Quick Reference

| Resource | Purpose | Example | |----------|---------|---------| | Pod | Smallest unit | Single container | | Deployment | Manage replicas | Web app | | Service | Network access | ClusterIP, LoadBalancer | | Ingress | HTTP routing | Path-based routing | | ConfigMap | Configuration | Environment variables | | Secret | Sensitive data | Credentials | | StatefulSet | Stateful apps | Databases |

Terraform Structure

project/
├── main.tf           # Resources
├── variables.tf      # Inputs
├── outputs.tf        # Outputs
├── providers.tf      # Provider config
├── versions.tf       # Version constraints
├── modules/
│   ├── vpc/
│   ├── eks/
│   └── rds/
└── environments/
    ├── dev.tfvars
    ├── staging.tfvars
    └── prod.tfvars

CI/CD Pipeline Template

# GitHub Actions
name: CI/CD
on:
  push:
    branches: [main]
jobs:
  build-test-deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Build
        run: docker build -t app .
      - name: Test
        run: docker run app pytest
      - name: Push
        run: docker push registry/app:${{ github.sha }}
      - name: Deploy
        if: github.ref == 'refs/heads/main'
        run: kubectl set image deployment/app app=registry/app:${{ github.sha }}

Monitoring Stack

┌─────────────────────────────────────────┐
│         OBSERVABILITY STACK              │
├─────────────────────────────────────────┤
│  Metrics:  Prometheus → Grafana         │
│  Logs:     Loki / ELK                   │
│  Traces:   Jaeger / Tempo               │
│  Alerts:   Alertmanager → PagerDuty     │
└─────────────────────────────────────────┘

Troubleshooting

Container not starting?
├─► docker logs <container>
├─► Check port conflicts
├─► Check image name/tag
└─► Check resource limits

Pod in CrashLoopBackOff?
├─► kubectl describe pod <name>
├─► kubectl logs <pod>
├─► Check resource limits
├─► Check probes configuration
└─► Check image pull secrets

Terraform apply fails?
├─► terraform plan first
├─► Check state lock
├─► terraform import existing
└─► Restore state from backup

High cloud bill?
├─► Enable cost alerts
├─► Right-size instances
├─► Use spot instances
├─► Delete unused resources
└─► Storage lifecycle policies

Common Failure Modes

| Symptom | Root Cause | Recovery | |---------|------------|----------| | Pod CrashLoopBackOff | App error or OOM | Check logs, increase limits | | ImagePullBackOff | Wrong image or auth | Verify image, check secrets | | Terraform drift | Manual changes | Import or terraform apply | | Slow deploys | Large images | Multi-stage builds, layer caching |

Best Practices

Docker

Use multi-stage builds
Run as non-root user
Use .dockerignore
Pin base image versions
Scan for vulnerabilities

Kubernetes

Set resource requests/limits
Use readiness/liveness probes
Store config in ConfigMaps
Use namespaces for isolation
Enable network policies

Terraform

Use remote state (S3, GCS)
Lock state file
Use modules for reuse
Plan before apply
Tag all resources

Next Actions

Specify your cloud platform and focus area for detailed guidance.

Agent Skills: Cloud Infrastructure Skill

Skill Files