🪨 IBM Z Xplore — PDS2: Advanced Filtering & Editing of Data Sets and Members

"Thankx for the Memberies" — IBM Z Xplore Challenge PDS2 · 250123-1116

Think of z/OS as the operating system that runs the computers powering your bank, airline reservations, and government records — the invisible backbone of modern civilization. It runs on IBM Mainframes, the most reliable machines ever built.

This challenge is about learning how to organize and process data inside that world-class system. Instead of regular computer folders and files, mainframes use something called Partitioned Data Sets (PDS) — a smarter, enterprise-grade way to store data that has been battle-tested for over 50 years.

In this project, I:

• 📂 Worked with PDS library members (think: files inside a folder, but on a mainframe)
• 📤 Uploaded data files into the z/OS file system through VS Code
• ⚙️ Ran a JCL job (a script that tells the mainframe what to do) called MERGSORT to merge and sort three rock data files
• 🏗️ Created my own custom Partitioned Data Set from scratch
• ✅ Validated the entire workflow using IBM's official checker job

Bottom line: I can navigate, organize, and automate data management on the world's most powerful and reliable computing platform — a skill set that's extremely rare and highly paid.

Your computer's folder system:
  📁 Documents/
    📄 resume.docx
    📄 cover_letter.docx

z/OS Partitioned Data Set (equivalent):
  📦 ZXXXXX.SOURCE          ← This is the PDS (the "folder")
    📄 ROCKS1                ← This is a Member (the "file")
    📄 ROCKS2
    📄 ROCKS3

The key difference: a PDS on z/OS has a directory that tracks exactly where each member is stored — allowing near-instant access on disks handling millions of transactions per second.

The diagram below shows exactly how data flows through this entire challenge — from raw text files on my laptop all the way to a validated output on the mainframe.

flowchart TD
    A[🖥️ Local Machine
rocks1.txt rocks2.txt rocks3.txt] -->|Upload via Zowe Explorer| B[📦 ZXXXXX.SOURCE PDS
Mainframe Storage]
    B -->|MERGSORT JCL Job| C[⚙️ JES Job Entry Subsystem
Job Scheduler]
    C -->|Reads INPUT datasets| D[🔀 MERGSORT Program
Merge and Sort Logic]
    D -->|Writes output| E[📦 ZXXXXX.OUTPUT PDS
ROCKSOUT Member]
    E -->|Rename member| F[✅ PDS2OUT Member
Final Output]
    F -->|Submit CHKAPDS2| G[🏆 IBM Validation
Challenge Complete]

    style A fill:#1a1a2e,color:#fff,stroke:#0f3460
    style B fill:#16213e,color:#fff,stroke:#0f3460
    style C fill:#0f3460,color:#fff,stroke:#533483
    style D fill:#533483,color:#fff,stroke:#e94560
    style E fill:#16213e,color:#fff,stroke:#0f3460
    style F fill:#1a1a2e,color:#00ff88,stroke:#00ff88
    style G fill:#0f3460,color:#ffd700,stroke:#ffd700

Downloaded three input data files to the local machine:

These files are the raw input that will be merged and sorted by the mainframe job.

Used Zowe Explorer inside VS Code to right-click on the SOURCE dataset and selected "Upload member…" to push all three rock files simultaneously to the mainframe.

💡 What just happened? Three plain text files traveled from a local laptop, across a secure network connection, into the file system of an IBM Mainframe running z/OS — ready for enterprise-grade processing.

Found the MERGSORT JCL job in the shared public dataset ZXP.PUBLIC.JCL and copied it into my personal JCL dataset.

//MERGSORT JOB ...
//STEP1   EXEC PGM=SORT
//SORTIN01 DD DSN=&SYSUID..ROCK1,DISP=SHR
//SORTIN02 DD DSN=&SYSUID..ROCK2,DISP=SHR
//SORTIN03 DD DSN=&SYSUID..ROCK3,DISP=SHR
//SORTOUT  DD DSN=&SYSUID..OUTPUT(ROCKSOUT),DISP=SHR
//SYSIN    DD *
  MERGE FORMAT=CH,FIELDS=(1,10,A)

The &SYSUID. variable is called a Symbolic. When JES processes the job, it automatically replaces &SYSUID. with the logged-in user's ID.

Before substitution:  &SYSUID..ROCK1
After substitution:   Z99999.ROCK1    ← Your actual dataset name

This is how one JCL script works for every user on the system — no manual editing required. This is exactly how enterprise automation works in banking, insurance, and government systems.

Before running anything, I analyzed line 8 of the JCL to predict the output dataset name — demonstrating an understanding of symbolic substitution. Then I submitted the MERGSORT job to JES.

sequenceDiagram
    participant Dev as 👨‍💻 Developer
    participant VSC as VS Code
    participant JES as JES Scheduler
    participant PGM as MERGSORT Program
    participant DS as z/OS Datasets

    Dev->>VSC: Submit MERGSORT job
    VSC->>JES: Job queued for execution
    JES->>DS: Read ROCK1, ROCK2, ROCK3
    DS-->>PGM: Stream input records
    PGM->>PGM: Merge and sort all records
    PGM->>DS: Write ROCKSOUT to OUTPUT PDS
    DS-->>JES: Job complete - RC=0000
    JES-->>VSC: Job output available
    VSC-->>Dev: ✅ ROCKSOUT member created

Created a brand new Partitioned Data Set from scratch using VS Code's Zowe Explorer:

1. Right-click on the connection profile
2. Select "Create New Data Set"
3. Named it using the required convention: ZXXXXX.PDS2.ANEWPDS
4. Selected "Partitioned Data Set: Default"
5. Clicked "Allocate Data Set"

ZXXXXX  .  PDS2  .  ANEWPDS
  ↑           ↑         ↑
HLQ         Level 2   Level 3
(Your UserID) qualifier  qualifier
              (max 8 chars each)

Refreshed the OUTPUT dataset view in Zowe Explorer, located the new ROCKSOUT member containing the merged and sorted rock data, then renamed it to PDS2OUT as required.

Cleaned up all temporary PDS2.* datasets created during the challenge, then submitted the official CHKAPDS2 validation job from ZXP.PUBLIC.JCL to have IBM's system confirm the challenge was completed correctly.

flowchart LR
    A[Submit CHKAPDS2] --> B{Validation Check}
    B -->|PDS2OUT exists?| C[✅ Pass]
    B -->|Temp datasets cleaned?| D[✅ Pass]
    B -->|Member renamed correctly?| E[✅ Pass]
    C --> F[🏆 Challenge Complete]
    D --> F
    E --> F

    style A fill:#0f3460,color:#fff
    style B fill:#533483,color:#fff
    style C fill:#006400,color:#fff
    style D fill:#006400,color:#fff
    style E fill:#006400,color:#fff
    style F fill:#B8860B,color:#fff

Here is how all the datasets relate to each other in this challenge:

graph TD
    subgraph LOCAL["💻 Local Machine"]
        R1[rocks1.txt]
        R2[rocks2.txt]
        R3[rocks3.txt]
    end

    subgraph MAINFRAME["🖥️ IBM Mainframe z/OS"]
        subgraph SRC["ZXXXXX.SOURCE PDS"]
            M1[ROCKS1 member]
            M2[ROCKS2 member]
            M3[ROCKS3 member]
        end

        subgraph JCL["ZXXXXX.JCL PDS"]
            J1[MERGSORT]
            J2[CHKAPDS2]
        end

        subgraph OUT["ZXXXXX.OUTPUT PDS"]
            O1[PDS2OUT member]
        end

        subgraph CUSTOM["ZXXXXX.PDS2.ANEWPDS"]
            C1[Custom Dataset]
        end
    end

    R1 -->|Upload| M1
    R2 -->|Upload| M2
    R3 -->|Upload| M3
    M1 -->|MERGSORT| O1
    M2 -->|MERGSORT| O1
    M3 -->|MERGSORT| O1
    J1 -->|Executes| O1
    J2 -->|Validates| O1

    style LOCAL fill:#1a1a2e,color:#fff
    style MAINFRAME fill:#0f3460,color:#fff
    style SRC fill:#16213e,color:#fff
    style JCL fill:#16213e,color:#fff
    style OUT fill:#006400,color:#fff
    style CUSTOM fill:#533483,color:#fff

Mainframes are the most secure computing systems in the world. Here's what makes this environment safe:

This challenge proves I can operate in environments that power the world's most critical infrastructure:

Less than 3% of developers in the world have any mainframe experience. Of those, very few have worked with both modern cloud tools AND z/OS simultaneously — which is exactly what this project demonstrates.

Global Developer Population:  ~27 million
  └── Have any mainframe experience:   ~800,000 (3%)
        └── Can use modern tools (Zowe/VSCode) with z/OS: ~150,000 (0.5%)
              └── Are actively building and shipping:  🫵 That's me.

📦 pds2-zos-advanced-datasets/
 ┣ 📂 images/
 ┃ ┣ 🖼️ vscode_dataset_menu.png     ← Zowe Explorer upload interface
 ┃ ┣ 🖼️ mergsort_jcl_code.png       ← Full MERGSORT JCL listing
 ┃ ┣ 🖼️ jcl_snippet.png             ← JCL symbolic variable example
 ┃ ┣ 🖼️ dataset_file_tree.png       ← z/OS dataset tree structure
 ┃ ┗ 🖼️ ibm_zxplore_logo.png        ← IBM Z Xplore branding
 ┣ 📄 README.md                      ← This file
 ┗ 📄 PDS2.pdf                       ← Original challenge document

This challenge is part of the IBM Z Xplore learning platform — IBM's official program for training the next generation of mainframe professionals. Completing the Advanced track demonstrates real hands-on capability with z/OS systems, not just theoretical knowledge.

IBM Z systems process:

• 🏦 Over $5 trillion in financial transactions daily
• ✈️ More than 95% of all airline reservations globally
• 🏥 87% of global credit card transactions
• 🏛️ Core systems for 71 of the top 100 banks in the world

If you're looking for someone who can build modern software and work confidently in the systems that keep civilization running — that's exactly what I bring to your team.