Migrating Mainframes to AWS with AWS Mainframe Modernization - Replatforming and Refactoring
Learn about migration patterns for mainframe applications using Mainframe Modernization, and when to choose Blu Age versus Micro Focus.
Overview of Mainframe Modernization
Mainframe Modernization is a service that supports the migration and modernization of mainframe COBOL/PL/I applications (often millions of lines of code) to AWS. It provides two approaches: replatforming (Micro Focus runtime) and refactoring (Blu Age automated conversion), enabling phased cloud migration of COBOL applications. The managed runtime environment executes mainframe batch jobs and online transactions. The service also includes migration assessment tools (Application Testing) that analyze mainframe workloads and generate migration complexity reports to determine whether replatforming or refactoring is more appropriate.
Replatforming and Refactoring
Replatforming runs COBOL source code without modification by building a Micro Focus runtime environment on AWS. Migration is fast (months-scale), and existing business logic is preserved as-is. The COBOL compiler and runtime operate in the cloud, so JCL, VSAM, CICS, and IMS interfaces remain usable directly. However, cloud-native scaling and serverless architecture cannot be achieved directly; only the execution platform changes while the mainframe structure is preserved. Refactoring uses Blu Age to analyze COBOL/PL/I source code and automatically convert it to Java. The converted application is Spring Boot-based and can be deployed to ECS or Lambda. VSAM files are mapped to RDS or DynamoDB, JCL batch jobs to Step Functions, and CICS transactions to API Gateway + Lambda. Refactoring requires a longer migration period (1-2 years scale) and greater investment, but after completion, you fully benefit from cloud-native capabilities (Auto Scaling, serverless, microservices decomposition).
Testing and Phased Migration
Run application tests in the Mainframe Modernization managed runtime environment to verify behavioral compatibility with the mainframe. Test both COBOL program batch jobs and online transactions (CICS equivalent), comparing output results with mainframe results. The Application Testing feature injects test datasets captured from the mainframe into the AWS environment and automatically detects output differences, enabling tens of thousands of regression test patterns to be executed without manual effort. In a phased migration, route some transactions to AWS while the mainframe continues handling the rest during a hybrid operation period. Data synchronization uses DMS with CDC (Change Data Capture) for real-time replication between the mainframe's DB2 and AWS RDS. To broaden your knowledge of migration projects, specialized books on Amazon can also be useful.
Selection Criteria - Replatforming vs. Refactoring
The decision framework rests on three axes: migration speed, depth of transformation, and cost structure. Replatforming is suitable when there are no plans to modify the COBOL assets and reducing mainframe operating costs is the top priority. For large enterprises where annual MIPS licensing costs run into hundreds of millions of dollars, migrating to AWS instance-based billing alone can yield significant cost savings. Refactoring is suitable when application feature expansion or microservices decomposition is planned, and maintaining COBOL is becoming a business constraint. A hybrid of both approaches is also effective: in practice, many projects keep stable batch processing running on the replatformed environment while refactoring only the frequently-modified customer-facing online processing into Java. An important consideration in the decision is COBOL developer availability. In organizations where COBOL engineers are declining due to retirement and aging, refactoring to Java broadens the pool of available maintenance personnel.
Pricing and Limitations
The replatforming managed runtime is billed by vCPU hours. The refactoring code conversion is billed by the number of converted code lines. Compared to mainframe MIPS (Million Instructions Per Second)-based licensing costs, AWS's pay-as-you-go model enables usage-based cost management. In the early stages of a migration project, build only development and test environments on AWS, then gradually scale down the mainframe after production migration to minimize the period of dual costs. As for limitations, Blu Age's automated conversion rate is generally 80-95%, with the remaining 5-20% requiring manual code completion or logic fixes. Programs that heavily use screen transitions (BMS maps) or proprietary macros tend to have lower conversion accuracy. Additionally, managed runtime region availability does not cover all global regions, so you should verify availability in your intended region beforehand.
Summary
Mainframe Modernization supports mainframe migration to AWS through two approaches: replatforming (Micro Focus runtime) and refactoring (Blu Age automated conversion). The managed runtime environment executes COBOL batch jobs and online transactions, while DMS CDC enables phased data synchronization with the mainframe.