Shell enterprise MDM foundation proof of concept disturbs rulebook

Shell has laid a foundation for enterprise master data management, choosing SAP MDM over Microsoft. The major lesson was that IT can build an enterprise MDM system to which the business will come.

Last year, oil and gas giant Shell executed an enterprise master data management (MDM) proof-of-concept project that ripped up the rule book.

It is a nostrum of corporate IT that IT should never lead the business. And so the drive for MDM should come from the business, not IT, and not data architecture. Yet Shell’s group lead data architect Andrew Schulze and Duncan Slater, an Accenture consultant working on the project, took the much derided “build it and they will come” path.

Schulze and Slater presented the proof-of-concept (PoC) project and its aftermath and adoption at IRM UK’s event this month -- Data Management, Information Quality, Data Warehouse & Business Intelligence.

Shell, which employs 93,000 people, uses the Data Management Association’s twofold definition of MDM: “processes that ensure that master reference data is kept up to date and coordinated across an enterprise. The organization, management and distribution of corporately adjudicated data with widespread use in the organization.” The PoC was intended to create the foundation of MDM both for business intelligence and to authorize master data back into source systems.

The project, which ran from November 2009 till March 2010, evaluated SAP MDM and Microsoft Master Data Services (MDS) against 16 MDM scenarios, from “match and merge” through “creating ERP business content” to security audit capabilities, but Schulze stressed that the proof-of-concept project was not a tool test.

“It was more ‘this is the business requirement from MDM; how can these tools help do that?’ ” he said.

Slater led a team for the project that was drawn from SAP, Microsoft, Accenture and Wipro. It was accountable to a steering committee consisting of architects and business representatives from the major business groups at Shell, and it met every Friday.

“The team makeup was a good blend of offshore and onshore,” Slater said, bringing together developers from Wipro’s Microsoft and SAP practices in India with Schulze and him, both South Africans. “That there was a test cricket series going on at the time added extra interest to the project!”

The team assessed the SAP and Microsoft MDM tools’ functionality and integration with Shell’s standard tools for workflow, data quality, messaging, Web services and extract, transform and load. Shell used Amazon Web Services for its hardware requirements, which made life much easier for the team.

“We did not want to be tied to procurement timelines that would eat into the budget, so we stood up a number of servers in Amazon Web Services,” Slater said. “That gave us a lot of flexibility and speed. It was quicker to get the hardware than the people.”

The PoC took 400 man-days, and the final choice was SAP MDM. “We felt that the Microsoft MDS tool showed a lot of promise and executed well across the scenarios,” Slater said, “but in terms of an enterprise class system -- regarding operations, audit, and so on -- and from an end-user friendliness perspective, it fell down somewhat.”

Since then, Shell has gone live in one part of the business, North American onshore gas, in its “field and reservoir group,” Schulze reported.

Underpinning the rollout across the company is a principle that Schulze described as “deploy by project, but build for the enterprise.” That is “easy to roll off the tongue, but is hard to do. The idea is that project efforts and resources, with a bit of extra governance and architecture, can be leveraged for everyone,” he said.

In the end, he said, “while business owns the data and its quality, we did [the MDM foundation project] from IT, but it works.”

Read more on Business intelligence and analytics

Data Center
Data Management