Background

Gen to hand-maintainable COBOL

An international finance organisation wanted to migrate its Gen estate to COBOL, so that it could be subsequently maintained by an outsourced development partner.

Overview

An international finance organisation wanted to migrate its Gen estate to COBOL, so that it could be subsequently maintained by an outsourced development partner.  The drive behind this migration was to:

Reduce dependence on declining skills base – Due to the reduction of the Gen skills base, resources were becoming more costly and difficult to scale.

Mitigate the risk of increased costs – Shared licenses for toolsets, generators, runtime licenses, etc. raised the risk of a financial impact on other areas of the business, if one part of the business migrated away.

Standardise technology – Using more recent, supported versions of technologies. The business mandated that all COBOL should conform to z/OS Enterprise COBOL standards.

The initial requirement was to perform a legacy application system upgrade which contained code from the 1990’s, last generated using Gen version 5.x (COOL:Gen) using a COBOL II compiler.

The subsequent migration project would then use the knowledge gained during this upgrade assessment and provide functionally equivalent, hand-maintainable code that had no dependencies on GEN runtimes.

Jumar delivered

  • Automated migration of Gen to COBOL using Jumar’s proprietary automation tooling. This enabled issues to be addressed once, then applied consistently to multiple in-scope screens
  • Resultant code that is functionally equivalent, performant, hand-maintainable and has no run-time dependency on Gen
  • Resolutions to model corruptions during the process
  • A solution aligned to the client’s development standards

Key benefits

  • Mitigation of risk – Due to the magnitude of this project and the client’s small number of in-house Gen specialists. Jumar’s team of worldwide experts, offered immediately available skills and technology.
  • Smooth transition to COBOL: Jumar’s expertise enabled it to analyse technical dependencies and versioning issues in a methodical way, so that the entire reference implementation was fully understood.
  • Savings: Migration away from legacy technology reduced the inherent costs associated with maintaining it.  Additionally, utilising automation tooling significantly reduced the time and cost involved with manual migrations.
  • Accelerated development: Jumar identified a common root cause to a wide-impacting issue and added support for the required fix to the code generator so that it only needed to be applied in a single location.
  • Easier IT maintenance: Removing the dependency on Gen (COOL:Gen) reduced the reliance on a dwindling skills base.  With the new smaller code base it is also easier to maintain.
  • Assisted strategy: Automation also allowed Jumar to migrate many out-of-scope Gen models into easily-to -understand COBOL, which aided the internal decision-making processes regarding the future strategy for those programs.

The project in more detail

Understanding the system:

The first stage in the engagement was to gain a full understanding of the source Gen (COOL:Gen) implementation, including a thorough analysis of a significant number of online 3270 screens and batches.

Approximately 300 screens were designated as in-scope for one of the early phases.  To accelerate this task, Jumar’s Model Analyser (part of the Jumar MAPS’ suite of automated software tooling) was used to provide an in-depth view of each screen, including complexity, Gen functionality and where technical dependencies resided.

This allowed Jumar and the client to analyse the screens with unprecedented accuracy from a number of perspectives, including: data driven analysis from the Gen model(s), action diagram logic, screen design and outside of Gen to create a priority list.

This process also identified screens which were no longer required. Such screens were removed from scope, reducing the effort required and simplifying the project, resulting in a much smaller code base.

Tailored automation:

Following the standard solution definition phase (where Jumar works with the client to specify the target solution, including: architecture, design patterns, standards, etc.), a phase of automation tailoring was factored in to address client-specific issues / standards that were identified.

Detailed technical dependencies for the entire solution were analysed and versioning issues identified. Determination of the correct version of master code was required as multiple potential versions of source code existed in certain cases – this prompted in-depth research into establishing the correct code base contents.

This is relatively straightforward for Gen (COOL:Gen) itself, but anything external with which Gen interacts (External Action Blocks, DCLGENs, Copybooks, etc), need to be managed carefully as they are artefacts outside the Gen model (and may, by their nature, be uncontrolled).

Migration from Gen to COBOL:

Facilitated by the earlier COOL:Gen upgrade project, Jumar’s Gen experts had a corruption free, ‘clean’ implementation that used the latest supported version of Gen as an excellent starting point for migration.  This allowed Jumar to use its XML Code Generation software, part of Jumar MAPS, to re-architect and re-write the in-scope Gen procedures (online 3270 green screen and batch) by taking objects through two automated re-write stages:

  1. Extraction into XML format
  2. From interim XML into COBOL, IMS-DC & DB2

Following completion of client QA and testing, Jumar successfully supported the deployment of the migrated application code into the production environment which met the requirement for functional equivalence, performance and hand-maintainability. Crucially, the resultant COBOL code was independent of the Gen runtime environment, achieving the key aim of removing the client’s reliance on the tool.

For more information about migrating away from Gen or about this case study, contact Jumar’s Gen specialists.

Share article: