Lam Bùi

Posted on Sep 23

Modernize Mainframe Applications with AWS Transform

#ai #aws #modernize

I. Concepts

1. What is AWS Transform?

AWS Transform is a new AWS service that uses agentic AI to automate modernization and migration of infrastructure, applications, and source code to AWS. It reduces the time, effort, and risk of handling legacy or complex systems such as mainframes, VMware, and .NET workloads by analyzing, refactoring, and redeploying on cloud.

2. Primary Use Cases

Mainframe Modernization

Transform COBOL, JCL, CICS, Db2, VSAM into modern Java applications.
Automatically analyze code, extract business logic, generate technical documents, and refactor.

VMware Migration

Migrate virtual machines from on‑premises VMware to Amazon EC2.
Auto-generate migration plans, configure VPC networking, and convert VMs to EC2.

.NET Modernization

Upgrade from legacy .NET Framework to cross‑platform .NET.
Integrates with Visual Studio; automatically resolves dependencies, builds, and tests.

Migration Assessment

Pre-migration cost and feasibility assessment.
AI-driven recommendations for EC2 sizing, BYOL optimization, and cost analysis.

3. Key Capabilities

Agentic AI with human-in-the-loop (HITL) for complex tasks and decisions.
End‑to‑end automation: discovery → planning → code refactor → deployment.
Flexible collaboration: role-based access (Admin, Approver, Contributor, Reader).
Workspaces to isolate projects, with dedicated jobs and connectors.
Worklog & dashboards to track progress and decisions.

4. Typical Workflow

1) Environment setup

Create an AWS account with administrative permissions.
Enable the AWS Transform service.
Configure AWS IAM Identity Center for users and groups.

2) Create a Workspace

Dedicated per transformation project.
Contains connectors, jobs, and artifacts.

3) Create a Connector

Connect to target environments (VMware account, GitHub repository, other AWS accounts).
Configure IAM policies and encryption with KMS.

4) Create a Job

Choose job type: VMware Migration, Mainframe Modernization, .NET Transform, etc.
Upload inputs (RVTools export, source code, JCL, spreadsheets, Git repo, etc.).

5) Execute Workflow steps

VMware: Discovery → Grouping → Generate VPC → Replicate & Migrate
Mainframe: Analyze COBOL → Generate Docs → Extract Logic → Refactor → Build Java
.NET: Connect Git → Resolve Dependencies → Transform Code → Test & Build
Assessment: Upload file → Cost estimation → Migration feasibility

6) Track Progress

Worklog: detailed action log.
Dashboard: overall job status and KPIs.

5. Security & Governance

Data encrypted by default using AWS KMS.
Option to use customer‑managed KMS keys for tighter control.
Comprehensive IAM policies for access control, connector creation, and deployments.
Supports cross‑region processing and compliance validation.

6. Roles

Role	Core Permissions
Admin	Full control: create workspace, job, connector; approve HITL steps
Approver	Near‑admin; cannot modify users, connectors, or workspaces
Contributor	Execute jobs, upload files, respond to HITL; cannot approve critical steps
Reader	Read‑only access to progress and artifacts

7. Highlights by Use Case

VMware Migration

AI‑generated AWS VPC topology mirroring on‑prem networks.
Automatic wave planning from discovery data.
Import from NSX or RVTools via collector or direct file upload.
Supports Hub‑and‑Spoke and Isolated VPC topologies.

Mainframe Modernization

Full support for COBOL, JCL, Db2, and VSAM.
Clear stages: analyze → extract logic → refactor to Java.
Re‑run per stage for iterative validation.
Outputs modern Java applications deployable on AWS.

.NET Modernization

Connect GitHub repository to retrieve code.
Automatically resolve missing libraries and dependencies; refactor to .NET 6/7/8.
Integrate with Visual Studio for review and edits.
Build and deploy to test environments.

8. Core Benefits

Accelerate migration/modernization by 50–80%.
Reduce manual errors on complex workloads.
Preserve critical business logic via intelligent extraction.
Scale for large organizations: multi‑account, multi‑region, IAM governance.

II. Step‑by‑Step Guides

1) Modernize Mainframe Applications with AWS Transform

1.1 Objectives

Use AWS Transform to modernize mainframe COBOL applications (commonly on IBM z/OS) by:

Analyzing COBOL/JCL/CICS code.
Generating technical documentation and extracting business logic.
Decomposing a monolith into functional domains.
Refactoring and transforming code into modern Java.
Planning migration waves.
Creating deployment pipelines on AWS.

The journey follows four phases: Assess → Mobilize → Migrate & Modernize → Operate & Optimize.

1.2 Procedure

Step 1: Prepare Source Code

Retrieve the demo repository CardDemo: https://github.com/aws-samples/aws-mainframe-modernization-carddemo
Select Code → Download ZIP.

After extracting, locate the app/ folder with multiple subfolders.

Step 2: Keep Only Required Folders

Keep these under app/:

✅ Keep	Reason
app/cbl/	COBOL programs
app/jcl/	Job Control Language scripts
app/cpy/	Copybooks (shared data definitions)
app/bms/	CICS screen maps (if present)
app/data/	EBCDIC data (test or demo)
app/proc/	JCL procedures and helper programs

Remove:

❌ Remove	Reason
catlg/, ctl/, csd/, cpy-bms/	Not essential for core analysis

Step 3: Zip the Source

Rename folder app to mainframe-src, then zip it.

Step 4: Upload to S3

Open AWS Console → S3: https://s3.console.aws.amazon.com/s3/
Create a bucket (e.g., lam-mainframe-transform).

Upload mainframe-src.zip to the bucket.

Step 5: Create Workspace and Job

Open the AWS Transform portal and sign in via IAM Identity Center.
Open your workspace or create a new one.
Select Create a job.

Choose job type: Mainframe Modernization.

To experience the full pipeline, select: Analyze code, Generate technical documentation, Decompose code, Plan migration sequence, and Transform code to Java.
Name the job and create it.

Step 6: Start Modernization

Skip Add collaborators and proceed to Connect to AWS account.
Enter your AWS Account ID and select Next.

In S3 console, copy the Bucket ARN of lam-mainframe-transform and paste it in AWS Transform to create a connector. Select Create connector.

Copy the verification link, open it in a new tab, and approve access to S3 for AWS Transform.

After admin approval, select Send to AWS Transform to continue.

Enter the S3 URI of mainframe-src.zip and select Send to AWS Transform.

Step 7: Analyze COBOL Code

When Analyze code completes, review results and select Send to AWS Transform to proceed.

Step 8: Generate Technical Documentation

Open Summary and select all folders in Selected Files.

Select Send to AWS Transform to begin Decompose code.

When complete, view the generated documentation in: s3://lam-mainframe-transform/transform-output/

Step 9: Decompose into Domains

Create a domain classification. Each domain represents a separate functional area.
Recommended initial domains and seed files:

Domain	Seed Files	Cyclomatic Complexity	Description (EN)	Reason for Selection
Customer Domain	CBCUS01C.cbl, COUSR00C.cbl	20 / 102	Customer and user information handling	Key customer/user management, moderate complexity
Transaction Domain	CBTRN01C.cbl, COTRN00C.cbl	59 / 100	Processing and managing financial transactions	Core transaction logic, high complexity
Card Domain	COCRDLIC.cbl, COCRDUPC.cbl	144 / 159	Card data listing and updates	Primary card ops, high complexity
Account Domain	CBACT01C.cbl, CBACT04C.cbl	1 / 98	Account-level data and operations	Foundational account logic, moderate complexity
Admin Domain	COADM01C.cbl	36	Administrative features and controls	Focused module, clearly separable
Common/Utility	CBSTM03A.cbl, CSUTLDTC.cbl	1 / 11	Shared utilities and helper functions	Commonly reused lightweight utilities