AWS Architecture Blog·October 1, 2026

Incremental Modernization of Legacy Airline Systems with AWS and Agentic AI

This article details Datalex's approach to modernizing its legacy Java EJB2 airline e-commerce system using AWS Experience-Based Acceleration (EBA) and agentic AI. The strategy focuses on incremental migration, leveraging a business service proxy and the Strangler Fig pattern to decouple and replace components without disrupting critical airline operations. Key architectural patterns for gradual modernization are highlighted.

Microservices Distributed Systems Cloud & Infrastructure

Read original on AWS Architecture Blog

The Challenge of Modernizing Core Legacy Systems

Datalex, a long-standing provider of airline e-commerce solutions, faced the common challenge of evolving a mission-critical, deeply integrated system built on a mature Java and EJB2 architecture. The goal was to modernize the runtime and delivery model to accelerate new feature development for 'Modern Airline Retailing' while ensuring zero disruption to existing, continuous airline operations. This scenario emphasizes the difficulty of transitioning from monolithic, tightly coupled systems to modern, agile architectures, especially when high availability and data integrity are paramount.

Incremental Modernization Strategy: Project Phoenix

Datalex's "Project Phoenix" adopted an incremental approach, utilizing an AWS Experience-Based Acceleration (EBA) workshop to prove feasibility and establish repeatable migration patterns. Instead of a 'big bang' rewrite, the strategy involved a phased modernization, with new Spring Boot microservices running alongside the existing n-tier architecture. This pattern is crucial for complex, high-stakes systems where a full rewrite is too risky or expensive.

Key Modernization Workstreams and Patterns

Microservice Prototyping: Extracted a slice of the Reservation component, refactored it into a Spring Boot microservice, proving technical feasibility and establishing patterns for code migration using AI-powered coding assistants.
DevSecOps Pipeline: Built an end-to-end CI/CD pipeline with Amazon ECS, ECR, and AWS Security Hub, embedding shift-left security practices for automated validation.
QA and Observability: Implemented comprehensive monitoring with CloudWatch, Datadog, and leveraged the Strangler Fig pattern with a gateway to route requests between existing and modernized APIs, enabling non-regression testing and real-time validation.
Agentic AI Proof of Concept: Demonstrated integration of generative AI using Amazon Bedrock AgentCore for a natural language interface, showing how new capabilities can be layered onto the system.

ℹ️

Architectural Decoupling with Strangler Fig and Proxy

A core architectural decision was the use of a business service proxy and the Strangler Fig pattern. The proxy intelligently routes incoming requests to either the legacy n-tier system or the newly deployed Spring Boot microservices based on migration status. This allows for individual services to be modernized, tested, and validated in isolation, while the legacy system continues to operate, ensuring backward compatibility and minimizing disruption during the transition.

Target Architecture Overview

The target architecture on AWS utilizes Amazon ECS for container orchestration. Kong API Gateway provides protocol translation (REST to SOAP) and dynamic routing. An agent orchestrator with Amazon Bedrock AgentCore manages specialized AI agents for authentication, data retrieval, and reporting. Apache Kafka forms an event-driven backbone for asynchronous communication, and a robust observability stack (CloudWatch, Grafana, X-Ray) provides real-time monitoring across all components. This layered approach demonstrates how to integrate modern cloud-native services and AI capabilities with existing proven systems.

Business Service Proxy: Routes traffic to existing or modernized services.
Modernized Microservices: Spring Boot services replacing EJB components on Amazon ECS.
Agent Orchestrator: Amazon Bedrock for AI-powered agents.
Event-Driven Messaging: Apache Kafka for asynchronous communication and decoupling.
Observability Stack: CloudWatch, Grafana, AWS X-Ray for comprehensive monitoring.
Security: AWS Secrets Manager, AWS IAM, and Kong API Gateway for secure access and API management.

Legacy ModernizationStrangler Fig PatternMicroservices ArchitectureAWSEJB MigrationCI/CDObservabilityAgentic AI

Comments

Loading comments...

Architecture Design

Design this yourself

Design a modernization strategy for a complex, mission-critical legacy e-commerce system (e.g., an airline ticketing platform) that must remain operational throughout the transition. Detail how you would use patterns like the Strangler Fig, introduce a business service proxy for incremental routing, and integrate new cloud-native microservices and AI-powered agents without downtime.

Practice Interview

Other design angles

· Design an API gateway layer that handles protocol translation (e.g., SOAP to REST) and intelligent routing between legacy and modernized microservices during a phased migration.· Architect an observability and CI/CD pipeline for a hybrid legacy-cloud environment to ensure continuous monitoring and secure, automated deployments during an incremental modernization.· Propose how to integrate AI-powered agents into an existing enterprise system to enhance functionality (e.g., natural language booking) while leveraging both legacy and new microservices.