←←All cases

SAP Data Services retirement for a European consumer business

Client

A consumer business inside a large European media group

Industry

Digital media

Services

Data engineering, ETL modernization, Gen AI-assisted data platform migration

Tech stack

SAP Data Services, SAP BusinessObjects, Apache Airflow, dbt, AWS, Snowflake, Python, OpenAI models

Challenge

Our client, a European company within a media holding that operates a diverse portfolio of lifestyle and entertainment platforms, had been using SAP Data Services for years to orchestrate, ingest, move, and transform data. Its ETL estate had become too complex and expensive to maintain, and the company already used a modern data tech stack (AWS-based Airflow and Snowflake). The client's goal was to migrate workloads from SAP Data Services to this environment and eventually phase out the use of this SAP product. The decision came as mainstream maintenance for Data Services 4.3 was set to end in 2026. Rather than invest in another supported version, the client chose to address the cost and complexity of its existing setup. Moving the jobs was more than a code conversion. Years of business logic and dependencies had to produce the same results in the new stack. The original and replacement workflows would need to run side by side before cutover, and the client’s data team had neither the spare capacity nor the Airflow experience for a lengthy manual migration. ITRex had previously worked with the client on its consumer platform. For this project, our task was to:

●

Analyze the SAP Data Services jobs, dependencies, and transformation logic

●

Move supported workloads to Airflow-managed pipelines in the client’s AWS environment

●

Compare the new pipelines’ outputs with the original jobs and resolve discrepancies

●

Keep existing workflows running during the transition

Solution

ITRex focused on the SAP Data Services ETL estate. A separate team handled the client’s migration of its SAP Business Objects to another reporting tool (Sigma BI).

A Gen AI-assisted conversion process

The initial SAP migration plan called for a manual lift and shift to maintain existing job logic. Instead of manually rebuilding each job, ITRex's data engineers created a Python-based accelerator (Endless Dev Tool) that uses OpenAI models to support conversion, validation, and error correction. The tool reads the SAP Data Services job repository XML exports, maps dependencies, categorizes workloads, and identifies conversion patterns. It then converts supported jobs into Airflow DAGs. The analysis also generates metadata that engineers can use to better understand the existing estate and identify issues that were difficult to detect in the original setup. The accelerator lived in the client’s AWS environment. OpenAI models worked with job definitions and structural metadata rather than production business data. Our team used a corporate OpenAI license that prevented model training on submitted information.

Automated validation & correction before cutover

The accelerator does not merely generate replacement code. Thanks to its self-healing logic, the tool compares the converted pipelines to the original SAP Data Services workflows, uses OpenAI models to diagnose discrepancies and revise the code, and then validates again. ITRex ran the old and new workflows in parallel so their outputs could be compared. The data checks took place within the client’s environment. The models could access business records, but the accelerator used them to analyze job structure, validate migration logic, and correct code—not to process the records themselves. Engineers reviewed the results and handled cases that required manual judgment. This repeated cycle of conversion, validation, and correction reduced the manual work needed to preserve the original job logic. It also gave the team evidence that the replacement workflows produced the expected results before retiring SAP Data Services.

Impact

ITRex migrated the client’s SAP Data Services workloads to Airflow-managed pipelines, bringing them into the AWS environment the company already operated. The engagement delivered:

●

3x faster migration. The Gen AI-assisted process reduced migration time by approximately three times compared with the planned manual approach.

●

A controlled transition. By running the old and new platforms concurrently while reconciling outputs, the data team could find and correct differences before switching to the new stack.

●

A simpler platform footprint. The migrated ETL workloads moved into the client’s existing Airflow environment, supporting its plan to retire SAP Data Services.

●

Code that the client can maintain. The client keeps the converted pipelines without taking on a recurring migration tool license.

The project also provided ITRex with a reusable method for analyzing SAP Data Services repositories, converting supported workloads, and validating the results. The level of automation in subsequent migrations will depend on the connectors, custom logic, dependencies, and target architecture.