Understand the shared responsibility model, core service categories, and how cloud billing works before architecting anything.
Cloud computing delivers on-demand IT resources over the internet.
Service models:
- IaaS (Infrastructure as a Service): VMs, storage, networking — you manage OS upward
- PaaS (Platform as a Service): managed runtime, you manage code/data
- SaaS (Software as a Service): fully managed app, you configure
- FaaS (Functions as a Service): event-driven, pay-per-invocation, zero server management
Deployment models:
- Public cloud: AWS, GCP, Azure — multi-tenant, shared infrastructure
- Private cloud: on-premises, OpenStack, VMware
- Hybrid cloud: mix of private + public with connectivity (Direct Connect, ExpressRoute)
- Multi-cloud: using AWS + GCP + Azure simultaneously for resilience or vendor lock-in avoidance
Shared Responsibility Model:
- AWS/GCP/Azure responsible for: physical data centres, hardware, global network, hypervisor
- Customer responsible for: OS patching, network config (VPC/SG), IAM, data encryption, app code
- In managed services (RDS, Lambda): AWS manages more, you manage less
Pricing models:
- On-demand: pay-as-you-go, no commitment, highest unit price
- Reserved Instances / Committed Use: 1-3 year commitment, 30-60% discount
- Spot / Preemptible: use spare capacity, 60-90% discount, can be interrupted
- Savings Plans: flexible compute commitment, applies across instance families
Service models:
- IaaS (Infrastructure as a Service): VMs, storage, networking — you manage OS upward
- PaaS (Platform as a Service): managed runtime, you manage code/data
- SaaS (Software as a Service): fully managed app, you configure
- FaaS (Functions as a Service): event-driven, pay-per-invocation, zero server management
Deployment models:
- Public cloud: AWS, GCP, Azure — multi-tenant, shared infrastructure
- Private cloud: on-premises, OpenStack, VMware
- Hybrid cloud: mix of private + public with connectivity (Direct Connect, ExpressRoute)
- Multi-cloud: using AWS + GCP + Azure simultaneously for resilience or vendor lock-in avoidance
Shared Responsibility Model:
- AWS/GCP/Azure responsible for: physical data centres, hardware, global network, hypervisor
- Customer responsible for: OS patching, network config (VPC/SG), IAM, data encryption, app code
- In managed services (RDS, Lambda): AWS manages more, you manage less
Pricing models:
- On-demand: pay-as-you-go, no commitment, highest unit price
- Reserved Instances / Committed Use: 1-3 year commitment, 30-60% discount
- Spot / Preemptible: use spare capacity, 60-90% discount, can be interrupted
- Savings Plans: flexible compute commitment, applies across instance families