Cloud Computing Basics
Quick Refresher for Advanced Developers
Since you already know cloud, this is a quick context section, not a tutorial.
Service Models
``` ┌──────────────────────────────────────────┐ │ SaaS: Applications (Salesforce, Office) │ ├──────────────────────────────────────────┤ │ PaaS: Platforms (PCF, Heroku, App Engine)│ ├──────────────────────────────────────────┤ │ IaaS: Infrastructure (AWS, Azure, GCP) │ ├──────────────────────────────────────────┤ │ On-Prem: You manage everything │ └──────────────────────────────────────────┘
Your responsibility ▼ ```
Where PCF Fits
| Layer | Provider | You Manage |
|---|---|---|
| IaaS | AWS/Azure/GCP | Applications, Container images, Config |
| PaaS | PCF, Heroku | Applications, Source code |
| SaaS | Salesforce, GitHub | Data |
Deployment Models
- Public Cloud: Shared infrastructure (AWS, Azure, GCP public regions)
- Private Cloud: Your own infrastructure (often PCF on-prem, customer-managed cf4k8s)
- Hybrid: Combination of public and private
Key Concepts You Already Know
Containerization
- PCF uses containers internally (via buildpacks)
- You don't write Dockerfiles for PCF
- Platform manages container lifecycle
Orchestration
- PCF has built-in orchestration (what Kubernetes does)
- cf4k8s uses Kubernetes as orchestrator
- You generally don't interact with Kubernetes directly
Multi-Tenancy
- PCF supports multiple orgs and spaces
- Similar to AWS accounts or cloud projects
- Quota and RBAC per organization
Auto-Scaling
- PCF can automatically scale instances
- Policy-based or manual
- Health checks for failed instances
What's Different in PCF?
Application-Centric Model
- In AWS: Think "I need 3 t2.medium instances with auto scaling"
- In PCF: Think "I have an app that needs instances"
- Platform handles instance type, networking, OS patching
Service Brokers
- Consistent way to provision any service
- Database credentials automatically injected
- No manual connection string management
Buildpacks
- Automatic language/framework detection
- Build process is standardized
- No need to create/maintain images
Deployment Comparison
AWS (IaaS)
git push → Triggers pipeline → Create AMI →
Launch instances → Configure load balancers →
Monitor + Scale → Done
Kubernetes (Container Platform)
git push → Build image → Registry → Create deployment →
Configure ingress → Configure HPA → Monitor → Done
PCF (PaaS)
git push → Done