As companies integrate more and more systems, both internally and externally, the reality of having one single version of data—master data management, in other words—becomes crucial.
SAP AG, Oracle Corp. and IBM are taking different approaches to MDM: Oracle with its customer data hubs, and IBM by buying functionality. Having spent the last year rebuilding its MDM functionality based on the acquisition of A2I Inc.s technology—and scrapping its initial plan to use SAP R/3s data model as the basis for an MDM offering—SAP is embarking on a fast and furious development cycle to bring its MDM capabilities to the fore.
SAPs goal is to service-enable its MDM tools as it completes its corporate ESA (Enterprise Services Architecture) road map and build out industry-specific data repositories, according to Nimish Mehta, senior vice president of NetWeaver enterprise information management at SAP, based in Walldorf, Germany.
To this end, SAP is concentrating on creating additional vertically oriented repositories for MDM. It already has a global data synchronization repository and is working on building out similar ones for high tech and oil and gas, to start. Eventually, it will look to build out more than 20 such MDM repositories.
The company also is releasing service packs every six months that bring not just incremental upgrades or fixes but actual functional upgrades. (Service Pack 4, scheduled for release next year, is next.)
In the process of implementing MySAP ERP 2004, along with SAP NetWeaver MDM 5.5, Andrea Pelosi, director of information management and data standards at Nortel Networks Ltd., said the service packs cant come fast enough.
"I would like SP4 to come out sooner," said Pelosi in Richardson, Texas. "We were hoping to have enhanced de-duplication with SP3, but it didnt come. ... I would like to see a lot of work there." The de-duplicate capability lets users compare product lists, for example, and get rid of duplicate entries.
SAP aims to develop an MDM tool that enables automatic data management across systems, according to Mehta.
"You should be able to take a custom data model that exists in legacy systems, and MDM should ... automatically be able to read that service call and ... adapt its own data model, without manual intervention," said Mehta. "That takes a lot of coding. Its basically an issue of granularity of service calls so that very specific transactions can be performed through service calls, versus the UI [user interface] as it is today."