Core lesson 3 of 18Topic: AWS foundations
On-premises vs AWS: infrastructure, operations and cost boundaries
On-premises and AWS differ in how resources are obtained and who manages each layer. Moving an application does not settle its data, access or operating responsibilities. Compare the same internal reporting application across equipment, OS, operations and cost, without assuming that cloud is always cheaper or fully managed.
Compare both options against the same requirements, identifying infrastructure responsibilities, retained operational work, cost components and migration checks.
Know cloud service models and distinguish OS, application and data. This lesson does not purchase equipment, enter contracts or create AWS resources.
Compare management layers for the same application
- Common requirements → On-premises optionAn on-premises arrangement may outsource work; examine its actual contracts.
- Common requirements → EC2-based optionThis branch uses EC2. A service such as S3 has a different management boundary.
- Common requirements
- Keep users, data, recovery needs and functionality comparable.
- On-premises option
- Review owned equipment, supplier contracts and operational staffing.
- EC2-based option
- AWS manages the infrastructure; the customer manages the guest OS and applications.
On-premises arrangements vary in location and outsourcing. AWS also offers different boundaries: an EC2 customer manages its guest OS, while other services abstract more of that work. “Inside or outside the office” is not enough to assign operational ownership.
Assign the work that remains after a move
| Area | On-premises checks | EC2-based AWS checks |
|---|---|---|
| Facilities and equipment | Ownership, facility provider and maintenance contracts | The infrastructure boundary provided by AWS |
| Guest OS and applications | Updates, configuration and operating owner | Customer-side updates, configuration and ownership |
| Data and access | Allowed users, operations, retention and recovery needs | The same needs, plus IAM and service-specific controls |
| Capacity and recovery | Expansion, maintenance, spare capacity and recovery process | Resource choice, limits, placement and recovery design |
Compare complete costs over the same period
| Component | Questions to include |
|---|---|
| Compute, storage and communication | Capacity, runtime, transfer destination, growth and fixed commitments |
| Equipment and contracts | Purchase, facilities, maintenance, refresh, connectivity and existing obligations |
| Operations | OS/application maintenance, incidents and staffing |
| Migration and exit | Data movement, parallel running, tests, cutover and reversal conditions |
| Retained resources | Unused resources, retained data and ongoing contractual charges |
Being able to change capacity does not guarantee lower total cost. Use current terms for the selected AWS services, Regions and purchasing options, and actual on-premises contracts and operating effort. This lesson supplies no fixed savings percentage or price.
Worked example: reports peak at month end
| Observation | Information still needed |
|---|---|
| Processing peaks at month end | Scaling unit, runtime, data volume and transfer time |
| The app reads an internal system | Connectivity, latency, identity and whether the dependency can change |
| It stores internal documents | Location, access, retention, deletion and recovery requirements |
| Equipment contracts remain active | Parallel-running period, exit terms, renewal timing and migration cost |
Choose an option to see the result and explanation.
Make the decision from requirements and evidence
- Write common requirements for users, data, function and disruption tolerance.
- Identify what moves and which dependencies, equipment and contracts remain.
- Assign management, updates and recovery for each layer.
- Compare costs over the same period, including migration and exit.
- Define a small evaluation, cutover conditions and a reversal plan before committing.
Check, then explain in your own words
- Separate responsibility for infrastructure, guest OS, applications and data.
- Compare complete costs under the same requirements.
- Explain why a month-end peak alone does not settle the choice.
Check your understanding: Does moving to AWS always remove guest-OS patching work?
It depends on the service. EC2 guest-OS management remains with the customer.
Use the service’s actual management boundary rather than location. Then assign the customer-side work to a concrete owner within the organization.
Sources and conditions checked
The official references used in this lesson are collected here with their scope, conditions and corresponding sections. Independent expert review and live AWS comparison have not been performed.
- AWS shared responsibility modelChecked:
Scope and conditionsResponsibilities vary by selected service and configuration
Used in this lessonCompare management layers for the same application Assign the work that remains after a move
Scope and conditionsSections 2–3: essential characteristics, service and deployment models. General definition, not current product pricing or a compliance determination.
Used in this lessonCompare management layers for the same application