Home / Blog / How Country Road Group built a pattern-driven integration…
Engineering

How Country Road Group built a pattern-driven integration solution on AWS

Country Road Group replaced BizTalk 2013 with a serverless AWS platform across 21 interfaces, resolving licensing debt and publishing 29 CDK patterns.

By Dillip Chowdary • Nov 20, 2026 • Source: AWS Architecture Blog

How Country Road Group built a pattern-driven integration solution on AWS
Image: AWS Architecture Blog

At a glance

Replaced middleware
Microsoft BizTalk Server 2013
New platform
Strategic Integration Platform on AWS
Active interfaces
21 retail integration interfaces
Retail footprint
5 brands across 600+ physical stores
Open-source library
29 AWS CDK integration pattern constructs
Compute limit
Sub-five-minute executions routed to Lambda

Country Road Group migrated its retail integration footprint from legacy middleware to a serverless Strategic Integration Platform on Amazon Web Services, according to AWS Architecture Blog's report. The Australian fashion retailer replaced Microsoft BizTalk Server 2013 across more than 600 stores and five apparel brands with a declarative, pattern-driven architecture. The implementation eliminates VMware hosting dependencies while establishing a unified integration model across twenty-one operational interfaces.

This report details the architectural layers, multi-lane routing mechanics, and operational tooling that power the Strategic Integration Platform. Enterprise architects, platform leads, and retail engineering teams will discover how cloud-native patterns can systematically replace costly enterprise service bus runtimes. It also examines the published open-source AWS Cloud Development Kit constructs that retail teams can adopt for equivalent integration modernizations.

01Legacy middleware debt and tooling consolidation

Country Road Group operated twenty-one integration interfaces connecting point-of-sale systems, Oracle RMS databases, and digital storefronts on Microsoft BizTalk Server 2013 hosted on VMware. The retailer faced looming end-of-support milestones, escalating VMware hypervisor expenses following Broadcom's acquisition, and a shrinking market talent pool of qualified BizTalk engineers. Maintaining proprietary adapters created substantial licensing and infrastructure costs without providing any new technical capabilities for retail operations.

The company also struggled with extensive tool sprawl, maintaining disjointed integration environments across BizTalk, Workato, Apollo, and Apache Kafka. Engineers lacked continuous delivery patterns and API-first mechanisms, which delayed partner onboarding across brands including Country Road, Mimco, Witchery, Politix, and Trenery. Platform engineers unified these separate environments into a single architecture requiring one shared skillset, accelerating deployment speeds while standardizing enterprise authentication and authorization protocols.

Security and compliance risks also forced a fundamental reassessment of dead-letter queue governance. Retail payloads often carry customer identities and transactional records, meaning failed integration records retain sensitive information. Country Road Group reclassified dead-letter queues as durable data stores, enforcing identical encryption, retention periods, and access boundaries as primary transmission channels to meet data residency rules.

02Three structural layers of platform architecture

The Strategic Integration Platform establishes a three-tier design encompassing a pattern layer, an AWS service layer, and an endpoint connectivity layer. The pattern layer defines twenty-one interfaces through standardized primitives such as publish-subscribe, content-based routing, and request-reply. Instead of writing custom orchestration code, developers declare integration patterns that translate directly into pre-built Terraform modules.

The AWS service layer coordinates serverless managed offerings shielded by AWS WAF and Amazon API Gateway entry points. Stateless event processing executes inside Spring Boot applications on AWS Lambda, while AWS Step Functions orchestrates multi-step flows with standardized error catches. Amazon EventBridge serves as the central event bus, assisted by Amazon SQS queues, Amazon SNS notification topics, and Amazon DynamoDB metadata storage.

The connectivity layer bridges on-premises installations, Oracle Cloud Infrastructure, legacy retail platforms, and cloud platforms like Salesforce Service Cloud and eStar Online. Core legacy engines include Oracle RMS, Manhattan Associates warehouse management systems, and Magenta Retail point-of-sale databases. File-based transmissions interface through AWS Transfer Family SFTP endpoints, while persistent payloads migrate to Amazon S3 Intelligent-Tiering and Amazon S3 Glacier storage classes.

03Consumer rate control and storage retention

Advertisement

Tech Pulse Daily

Get tomorrow's pulse first

Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.

Country Road Group designed the Strategic Integration Platform around consumer-driven data retrieval rather than accepting unbounded producer pushes. By actively pulling data from source systems on predetermined schedules, the platform controls intake velocity and prevents downstream retail systems from experiencing traffic saturation. The architecture ensures that retail databases and warehouse endpoints process messages within safe throughput boundaries.

Note: SIP retrieves data from source systems on its own schedule rather than accepting unbounded pushes, allowing the consumer to control the intake rate. When SaaS webhooks force a push, the platform catches incoming payloads in a queue before processing them at a controlled pace.

Persisted data receives automated lifecycle management to satisfy regulatory retention schedules and reduce ongoing storage overhead. Payloads flowing through Amazon S3 transition between access tiers based on age and retrieval frequency. Storage policies route older transactional archives directly into Amazon S3 Glacier vaults while preserving queryable indexing in Amazon DynamoDB.

04Four pattern-led service bus lanes

The service bus splits message traffic into four distinct processing lanes aligned with payload sizes and processing duration. Lane one manages batch managed file transfers, using Automic or EventBridge triggers with AWS Transfer Family to archive files in Amazon S3 before triggering Lambda parser functions. Lane two handles synchronous REST calls through Amazon API Gateway, passing payloads to Step Functions and an automated transformation router.

Lane three facilitates direct database connectivity for bulk data extraction using AWS Lambda and Amazon AppFlow for lightweight tasks, or containerized workers for heavier queries. Lane four routes asynchronous event traffic, receiving external requests via API Gateway and internal updates through EventBridge before broadcasting over Amazon SNS and queuing inside Amazon SQS.

Workload characteristics dictate whether transformations execute in serverless functions or containerized services. Short stateless transformations run on Spring Boot Lambdas within five-minute execution windows, while extended stateful jobs delegate to Apache Camel on Amazon ECS with AWS Fargate.

| Compute Runtime | Assigned Workload | Execution Duration | | AWS Lambda | Stateless transformations, parsing, API workflows | Sub-five-minute executions | | Apache Camel on ECS Fargate | Stateful integrations, large files, legacy protocols | Long-running executions |

05Unified error management and operational recovery

Failed transactions across all four lanes divert into an Amazon SQS dead-letter queue governed by the custom Error Management System. The system extracts payload bodies into Amazon S3 and indexes error metadata in Amazon DynamoDB, tagging each failure with a unique correlation identifier and standard JSON contract. The backend operates on Spring Boot within AWS Lambda, complemented by an AWS Amplify Next.js dashboard.

The visual dashboard enables first-level and second-level support operators to investigate, adjust, and resubmit failed integration transactions idempotently without software engineering intervention. Every newly created interface connects to error monitoring automatically through a designated interface identifier. This self-service operational tooling eliminated previous level-four developer triage escalations, establishing faster incident recovery and higher production uptime.

To assist other organizations facing legacy middleware migrations, the implementation team published the Enterprise Integration Patterns AWS Cloud Development Kit Library as open source. The release contains twenty-nine pre-built constructs implementing standard enterprise patterns as reusable cloud infrastructure. The package also features an agent skill that assists artificial intelligence coding assistants including Claude Code and Kiro in composing and configuring patterns.

Frequently asked questions

Why did Country Road Group decommission Microsoft BizTalk?

Country Road Group decommissioned BizTalk 2013 because it was approaching end-of-support, incurred rising VMware costs under Broadcom, and faced a vanishing talent market of BizTalk developers.

How does the Strategic Integration Platform handle transaction errors?

Failed transactions route to an Amazon SQS dead-letter queue, store payloads in Amazon S3, and index metadata in Amazon DynamoDB with a correlation identifier. Support operators inspect, edit, and replay failed messages through an AWS Amplify Next.js dashboard.

When does the architecture use ECS Fargate instead of AWS Lambda?

AWS Lambda handles stateless event processing and short jobs completing in under five minutes. Apache Camel hosted on Amazon ECS with AWS Fargate takes over long-running, stateful, and large-file integration flows.

What open-source tools were released from this project?

The team released the Enterprise Integration Patterns AWS CDK Library, containing 29 reusable infrastructure constructs. The repository includes an AI agent skill that helps assistants like Claude Code and Kiro configure patterns.

Sources

Dillip Chowdary

Author

Dillip Chowdary

Writes Tech Bytes coverage of AI, engineering, and the tools that actually ship. Editor of Tech Pulse Daily.

Related on Tech Bytes

Advertisement

5-min tech signal

Weekday briefing for engineers who skip the noise.

No spam · Unsubscribe anytime

Advertisement

✈️ CareerPilot

Your AI job-search copilot

Match your resume against live Ashby, Greenhouse & Lever openings — fit scores, job-specific resume optimization and email alerts.

Find matching jobs →

Free Tools

Browse all tools →