Visual Process Automation (Batch & Atomic Fan-Out)

Overview

This use case demonstrates MdlWr's advanced Visual Process Automation capabilities, specifically highlighting how asynchronous jobs handle bulk data items.

In enterprise workflows, a single incoming payload often contains multiple records (an array of items). MdlWr allows you to process these items flexibly through connected nodes:

• Batch Mode (1 Job Triggers 1 Job): Bundles all items together and passes them downstream as a single array payload.
• Atomic Mode (1 Row Triggers 1 Job / Fan-Out): Automatically splits (fans out) the array, spawning an individual, independent asynchronous job for every single row/item.

---

Prerequisites & Architecture Workflow

[ Incoming Payload (Array of Items) ]
│
▼
[ MdlWr Async Job ]
│
├── (Option A: Batch Mode) ──> Sends entire array as 1 bundled payload to Next Job
│
└── (Option B: Atomic Mode) ──> Fans out into N separate jobs for each individual row/item
│
├── Item 1 ──> [ Node 1: Hello World ] ──> [ Node 2: Random Complete Time ]
├── Item 2 ──> [ Node 1: Hello World ] ──> [ Node 2: Random Complete Time ]
└── Item N ──> [ Node 1: Hello World ] ──> [ Node 2: Random Complete Time ]

Follow the step-by-step implementation guide below to configure and test batch and atomic processing flows within the platform.

---

Step-by-Step Implementation Guide

Step 1: Create a New Logic Flow

Set up the foundational asynchronous workflow to receive bulk data payloads.

1. Navigate to Integration > Logic Flows.
2. Create a new logic flow and configure its processing mode to Async and execution mode to Webhook.
3. Fill in required metadata such as Name, Path, and Category, then save.

Step 2: Add Initial API Node (Hello World)

Incorporate the first processing block into your visual pipeline.

1. Open the logic flow visual canvas/editor.
2. Drag and drop the built-in sample API node: MDLWR > Sample > Sample - Hello World.

Step 3: Add Subsequent API Node (Random Complete Time)

Incorporate the next step in the pipeline that will receive data from the previous node.

1. Drag and drop the second sample API node: MDLWR > Sample > SAMPLE - Run Random Complete Time.

Step 4: Connect the Nodes

Establish the execution pipeline path.

1. Connect the output of the Hello World node to the input of the Random Complete Time node.
2. Ensure the connection is tied to the Success execution branch.

Step 5: Configure Connection Logic (Batch vs. Atomic)

Determine how data passes from one node to the next by editing the connector rules.

1. Right-click on the connecting line between the two nodes.
2. Select Edit Logic from the context menu.

3. Choose your preferred processing mode:

   • Batch Mode: Select 1 Job Triggers 1 Job. This bundles and sends all incoming data items downstream as a single array.
   • Atomic Mode: Select 1 Row Triggers 1 Job. This activates Fan-out mode, breaking the array down so every individual row triggers its own isolated downstream job.

Step 6: Apply Python Filtering Logic (Optional)

Add conditional execution rules to filter rows dynamically during processing.

1. Within the connection logic settings, insert custom Python evaluation code if needed.
2. For example, use a condition like return row_data['status'] == 'Active' to ensure only matching rows proceed to the next node.
3. Save the connection configuration.

Step 7: Save the Logic Flow

1. Click Submit to store the complete visual process flow configuration.

---

Testing via Flow Tester

Verify how your batch or atomic fan-out logic behaves under live payload conditions.

1. Navigate to System > Flow Tester.
2. Select your newly created visual process automation flow using the folder browser.
3. Provide a test payload structured as an array of items (e.g., simulating multiple user or transaction records):

{
 "items": [
   {"id": 1, "name": "Alpha Corp", "status": "Active"},
   {"id": 2, "name": "Beta LLC", "status": "Pending"},
   {"id": 3, "name": "Gamma Inc", "status": "Active"}
 ]
}