Skip to main content
BUSINESS CHANGE REQUEST

Incremental Loading Implementation

The Customer Master pipeline currently performs a full data load every night. Due to rapid business growth, execution time has increased significantly, impacting SLA compliance and increasing infrastructure costs. The Data Engineering team has been asked to enhance the existing pipeline by implementing an incremental loading strategy that processes only newly created and modified customer records.

BUSINESS JUSTIFICATION

Why This Enhancement Matters

The existing Customer Master pipeline successfully delivers accurate data, however its full-load implementation has become increasingly expensive and time-consuming as customer volume has grown. Leadership has approved this enhancement to improve operational efficiency while maintaining the same business functionality.

Current Runtime
96 min
Nightly Full Load
Target Runtime
< 25 min
Incremental Load Goal
Daily Records
480 M
Customer Records Processed
Daily Compute Cost
$1,450
Current Infrastructure Spend
CURRENT PIPELINE OVERVIEW

Existing Customer Master Pipeline

Before implementing incremental loading, the engineering team reviewed the existing production pipeline to understand its current architecture, execution flow, and processing pattern. The pipeline is stable and functionally correct, but it reloads the complete customer dataset every night.

Source System
CRM Database
Extracts the complete customer table every night using a scheduled batch process.
Bronze Layer
Delta Lake
Stores the raw customer dataset exactly as received from the source system.
Silver Layer
PySpark Transformations
Applies cleansing, standardization and business rules before publishing curated data.
Gold Layer
Snowflake
Publishes customer data for dashboards, reporting and downstream business applications.
Skip to main content
Skip to main content