Reference solutions/Reference architecture
Reference architecture

A web platform with room to grow.

A reference pattern for business websites, customer portals and online commerce that need clear network boundaries, scalable application capacity and observable operations.

The workload question

Resilient web platform

You need to separate public delivery from application and database tiers, scale for changing demand and make deployments easier to operate.

ILLUSTRATIVE DESIGN

This pattern illustrates component relationships. Region availability, data location, application constraints and the final service choices are assessed before implementation.

Related delivery service
AWS REFERENCE DESIGN
Amazon CloudFrontContent delivery
→
AWS WAFEdge protection
ILLUSTRATIVE WORKLOAD ENVIRONMENT
Application Load Balancer
AVAILABILITY ZONE A
AWS FargateApplication tasks
AVAILABILITY ZONE B
AWS FargateApplication tasks
Amazon RDSMulti-AZ database
Amazon S3Object storage
Edge delivery, application capacity across two Availability Zones, and managed data services.
Illustrative relationships. Final configuration follows workload discovery.
Design decisions

The choices behind the components.

Architecture is a set of trade-offs. These are the questions to resolve before implementation.

01

Availability

Distribute application tasks across Availability Zones. Assess database Multi-AZ options and test health-check and failover behaviour.

02

Data protection

Keep database access private, scope S3 permissions and encryption, and align retention to the business requirement.

03

Delivery

Build and scan images, deploy through an approved pipeline and use application health checks before accepting a release.

04

Cost

Estimate a baseline, peak demand, data transfer and service charges. Scale the pattern to the expected traffic rather than assuming every component is required.

Validation before handover

Test the assumptions.

The acceptance plan should check the behaviour of the full workload, alongside individual AWS resources.

  • Test application, database and edge-cache behaviour with representative requests.
  • Exercise a failed task and an agreed database recovery scenario.
  • Confirm alerts and release rollback meet the operating requirements.

What a project adds

A workload-specific architecture, implementation plan, cost assumptions, operating runbooks and a record of agreed validation checks.

Your next step

Explore this pattern
for your workload.

Tell us about your workload, current challenges and timing. We’ll use that context to discuss a suitable scope.

Email Sunrise