Skip to content

"Honey! I shrunk the..." mainframe modernization effort

10 minute read
Content level: Intermediate
1

What are you talking about?

✨Mainframe... & Modernization...

Mainframes are large centralized computers used by organizations to support mission critical applications and designed to support high performance batch and online transaction processing and are used across industries.

IBM Z dominates 85-90% of the mainframe market, while other mainframe still in use are Unisys ClearPath MCP & OS2200, Fujitsu BS2000 & VME, and Bull GCOS. Midrange platforms such as IBM i (AS/400), HP NonStop, OpenVMS, IBM AIX, HP-UX, Oracle Solaris are often referred to as mainframes.

Mainframes are used today across industries in various lines-of-business:

Enter image description here


Enter image description here

🤔 Mainframes are great machines, then what's the problem?


Enter image description here

Organizations with business critical applications running on mainframes are facing challenges related to high costs, lack of agility and risks associated with decades of technical debt and shortage of legacy skills.

Enter image description here

Mainframe workloads are typically monolithic in nature and contain decades of technical debt due to outdated technical components and a lack of documentation, causing negative impacts on release cycles, maintenance costs, and business outcomes.

🧐 Okay, so there are benefits...


Absolutely !!

Enter image description here

💰 Cost Savings:

A key driver for moving off the mainframe is cost savings, however organizations can’t base the entire justification for migrating on cost alone. One can reduce the mainframe cost by reducing the consumption and migrating your mainframe applications to Cloud which offers the pay-as-you-go model. The other factor that significantly helps to reduce the cost is moving away from proprietary ISV software to open-source stack. Looking at completed migration projects, we see organizations achieving a 60%-90% cost reduction once migrated to AWS.

🛡 Reduce Risks:

Organizations want to avoid the risks of very slow mainframe changes because of complexity, archaic technology and using a platform in a silo. They want to address technical debt and avoid vendor-lock in. Many organizations have challenges with legacy skills. Mainframe specialists are retiring, taking their expertise with them. It’s becoming prohibitively expensive to find personnel with the requisite knowledge. Migrating to cloud organizations can solve the retirement skill gap and attract new talent with cutting-edge skills.

🎢 Business Agility:

Organizations want to accelerate the ability to meet their customer needs, and increase agility and productivity via access to native cloud services. Mainframes typically host critical applications with core-business data. Organizations are looking for ways to unlock the mainframe data stores in order to exploit data analytics, machine learning and artificial intelligence.

🚀 Innovation:

Once legacy applications migrated to AWS, they can be integrated with 200+ other services including advanced analytics, machine learning, generative AI, etc. Companies can quickly spin up an environment to test their ideas on-the-fly. Those options are not just available when on mainframe.

🤷 What's holding them back ?


Well !!

Mainframe migrations are lengthy 🧭 and costly 💵. There should be a driving factor like the ones mentioned above. Also, organizations have to understand and layout the plan and approach of the complete migration process before they can embark the journey.

Migration of code and data is generally 30%-40% of the migration effort. Organizations have to plan around the peripheral activities related to mainframe migration. This includes setting up security, configuring scheduler for batch, considering the non-functional requirements, synchronizing the changes with external services or trading partners with whom your mainframe exchanges data, and most significantly the testing to make sure your target migrated application is producing the expected results.

Enter image description here

There are a number of patterns (often called the "R's") that companies can follow to modernize mainframe workloads and migrate to the cloud. It’s not a one size fits all approach, and companies often develop strategic plans which include a combination of patterns.

THE '8 R APPROACH'

Enter image description here

✳️ 1. Reuse:

Enhance existing mainframe application with integrations and cloud innovations
A data augmentation approach allows customers to unlock their legacy data and leverage AWS cloud-native services. This pattern supports near real-time replication of data for analytics, AI, machine learning, and enables new channels and new function use cases.
When to use: Customer has urgent drivers for bringing data to AWS. Customer is looking to lower mainframe consumption costs by moving data to AWS. Often the quickest path to value and the first step for many companies.

✳️ 2. Reimagine:

Redevelop/carve application in new language & cloud native architecture
Rewrite or rebuild applications to leverage cloud-native architectures such as Java microservices, and AWS services such as AWS Lambda and AWS Glue. Companies are beginning to leverage generative AI to accelerate these projects. Tooling such as AWS Transform for mainframe provides the ability to analyze code, understand business rules, generate documentation, decompose monoliths, plan the migration sequence and assist with code transformation.
When to use: Application no longer meets the needs of the business. Customer doesn't want to retain any legacy technical debt in the target environment. Customer is often targeting low-code platforms such as Appian or Salesforce Lightning.

✳️ 3. Refactor:

Transform application code & data to new languages and modern architectures
Retains functional equivalence while converting core legacy applications to cloud-native language such as Java or .Net code. Once legacy application and database is converted, developers can extend application functionality directly without having to navigate the original legacy code. Proven automated Refactor solutions include AWS Blu Age, mLogica, Deloitte InnoWake, IBM ModernSystems, TSRI, Astadia Anubex and Ensono.
When to use: Core application functionality meeting business needs however application development skills becoming an issue. Cost savings is a driver, and there’s an aggressive timeframe to complete the project. Mandated target environment is “legacy free”.

✳️ 4. Replatform:

Port & recompile application to a modernized cloud runtime environment
Move mainframe workloads to AWS without changing the underlying technologies. Proven, mature modernization approach with first project delivered in early '90s. Proven Replatform tools and frameworks include the Rocket Software (was Micro Focus) Enterprise Server, NTT DATA UniKix, TmaxSoft, IBM ModernSystems and Heirloom.
When to use: Core application functionality is meeting business needs and application development skills not an issue. Cost savings is primary driver and there’s an aggressive timeframe to complete the project. Also an option if the application is close to end of life and will be retired or replaced in the upcoming years – however has become an obstacle to exiting the mainframe.

✳️ 5. Repurchase:

Purchase and customize COTS software package to replace application
Replace existing applications with a commercial off the shelf packages. SAP and Oracle ERP are good examples of COTS solutions that are widely adopted.
When to use: An issue is when the existing legacy applications have been heavily customized and not supported by the target package. Applying an additional pattern such as Refactoring can look to address this functionality gap.

✳️ 6. Rehost:

Redeploy application & dependencies unchanged to other infrastructure
For the IBM Z mainframe, customers may choose to Rehost (or Relocate) their mainframe to a hosting partner such as Kyndryl or Ensono, then perform an incremental modernization to AWS. There are also rehost options for the following platforms: IBM i (AS/400), IBM AIX, Oracle Sparc Solaris and Unisys ClearPath. The latter two run emulative solutions on AWS.
When to use: Customer needs to exit the data center quickly, e.g., impending contract renewal or end of support platform. And/or one of the systems is a blocker in a larger data center exit.

✳️ 7. Retain:

Keep application unchanged when no business case for modernization
Applications are retained on the mainframe without modification. Possibly look to replicate data (see Reuse).
When to use: Applications that you’re not ready to modernize, or that still meet the needs of the business with infrequent changes required. Alternatively, the applications might be scheduled to be retired in the near future, therefore you can simply retain them and do nothing for now.

✳️ 8. Retire:

Decommission application including its code and data stores
Some applications might not be meeting business requirements and/or are no longer needed, so can therefore be decommissioned and retired.
When to use: Applications are no longer meeting business needs can be retired. Any required feature functionality can be replaced by another existing or planned new system.

🤨 Can't Generative AI help to modernize the mainframe applications?


Now we are talking...

Early adoption of Generative AI is showing significant benefits. Modernization of mainframe is a daunting task, with high cost and stretched timeline. Generative AI is shortening the timeline and reducing the cost drastically. What earlier used to take months or years are now completed in days and weeks.

Timeline for Traditional migration

Enter image description here

Timeline for AI assisted migration

Enter image description here

Generative AI can perform the modernization tasks with ease and accuracy: 

  • Automating documentation of legacy code bases
  • Extracting business rules from complex systems
  • Generation of test cases and test data
  • Accelerating new application development

Generative AI proves most valuable in "Reimagine" modernization scenarios, where systems undergo complete transformation. Yet for existing "Refactor" and "Replatform" patterns, where functional equivalence is critical, Generative AI serves as a powerful complement. When integrated with proven tooling, Generative AI enhances efficiency by: 

  • Identifying optimal program and data groupings for phased implementations (go-live wave planning)
  • Generating comprehensive test cases and test data to accelerate the overall test phase
  • Aiding with the resolution of missing or incomplete code bases
  • Optimizing transformed code quality and performance

AWS Transform for mainframe is first agentic AI service for large-scale migration and modernization of mainframe application.

Enter image description here


AWS Transform for mainframe:

  • Analyzes millions of lines of code for comprehensive codebase understanding that eliminates modernization surprises by identifying risks, dependencies, and complexity upfront
  • Analyzes data structures of data sources to build complete relationship visibility that maps program-to-data connections and operation types
  • Combines static analysis results with execution data gathered from the mainframe via System Management Facility (SMF) to produce component activity metrics and identify unused jobs and transactions
  • Generates documentation at both technical and business levels with comprehensive end-to-end process visibility, process flows, and business rules
  • Decomposes monolithic mainframe applications into manageable, business-aligned domains and provides suggestions for migration planning sequence
  • Enables users to chat in natural language with the AI agent about business rules, technical documentation, SMF activity analysis, dependency graphs, data lineage, data dictionary, and scheduler
  • Refactors millions of lines of COBOL code to Java/Angular in minutes preserving exact business functionalities
  • Generates automated test plans, data collection scripts, and execution automation based on actual application dependencies and business logic

That's what I was talking about !!!

Generative AI's role in mainframe modernization is rapidly evolving, with early adopters already seeing promising results. As the technology matures, its transformative potential becomes increasingly clear - from accelerating cloud migration initiatives to significantly reducing cost and risk across the modernization journey.


👀 I have some questions...


I know...I know...Let's ask in re:Post Questions.


If you want to read more about Reimagine of mainframe application, check the blog and article

Parts of this content courtesy of Barry Tait.