Central Data Delivery Platform

A company with multiple applications faced a classic problem of distributed system landscapes: Each application had built and managed its own staging layer for ERP data (SAP and other systems). The result was many different formats, inconsistent data and high maintenance effort across distributed teams.

Requirement

The requirement was a central data platform that prepares ERP data once and provides it for an entire ecosystem of applications – for both transactional and analytical workloads. At the same time, cross-cutting analyses on ERP data combined with application data should be enabled.

Architecture

The developed architecture uses Microsoft Fabric as the central platform. Sync processes replicate ERP data to Fabric, where they are prepared nearly-real-time via Materialized Lake Views for both transactional and analytical processes. The Materialized Views enable performant queries without the complexity of classic ETL pipelines for each consumer.

Result

The concept eliminates redundant staging layers: Applications consume directly from the central platform instead of maintaining their own copies. For transactional applications, GraphQL endpoints were provided that allow querying only the actually needed fields – efficient and flexible.