Research Best Practices
This guide collects workflows that work well for business analysts and architects using Zengines Data Lineage. Developers can use the same paths when validating logic for migration or rewrite scope.
If COBOL or RPG terminology is new to you, read COBOL Basics for Lineage Research or RPG Basics for Lineage Research first.
Start from edges, not isolated fields
Within a module—and across called modules when they are in the project—trace edge to edge:
- Find an input edge (table read, file read).
- Follow links to variables and the top variable / record structure.
- Walk transforms through paragraphs, subroutines, or related children.
- Find the matching output edge (table write/update, file out).
You are usually researching a group (a table row, a file record, a business outcome), not one elementary item alone.
See Tracing Data Lineage in Explore.
Choose the right front door
|
You know… |
Start here |
|---|---|
|
Business question, not program names |
Dashboard → Ask a Lineage Question |
|
Table or column names |
Fields tab (filter by table) or Tables tab |
|
Program name |
Modules tab |
|
Job or data set name |
Jobs or Files tab |
|
Variable name (usually from a Zengines spreadsheet Link) |
Variables tab |
Fields-first for data model experts
Many analysts prefer the Fields tab:
- Filter to one table, adding a module filter if applicable.
- Review every column the module touches.
- Open a field → jump to module and variable for lineage detail.
This is often faster than opening programs at random when the data model is already familiar.
Modules-first for program experts
When you know the COBOL name:
- Open Modules → select the program.
- Review table and file edges on the detail view.
- Sort variables by child count to find record-level working storage.
- Link into Variables for paragraph- and statement-level detail.
Batch and file research
When data moves on files:
- Jobs tab → filter by module if you know the program.
- Open the job → follow variable links for record layouts.
- Use Files tab to see all jobs and modules for a data set name.
Include job language sources (JCL, CL, CLLE, and related) in ingestion for the richest file graph. See Understanding your Data Lineage Project.
Using AI without skipping proof
|
Tool |
Best use |
|---|---|
|
Ask a Lineage Question |
Discovery and orientation; open module links in a new tab to keep the answer visible |
|
Summary |
Quick business read of one module, variable, or job |
|
Tell Me More |
Deeper AI summary; click links to verify in Explore |
|
Explore links |
Verification and audit-ready paths |
Ask specific questions when you can (“overdraft fees”, a module name, a table). Broad questions work; narrow follow-ups save time in large projects.
Recently asked questions on the dashboard is local to your browser.
Sample project examples
Zengines sample projects use generic banking-style COBOL (accounts, retail posting, fees). Example prompts that match training scenarios:
- “Tell me about overdraft fees” → often one computing module
- “Does the retail summary program interact with any files?” → module-scoped answer with file/table links
Use sample projects to practice edge-to-edge tracing before your production project is loaded.