A prospective employee walks onto a job site and applies for a job. How much will that employer know about the applicant?Under immigration reform legislation, a stipulation called the Employment Eligibility Verification System would mandate that employers check the eligibility of all current or potential employees against federal and state databases.Potentially, information, or lack of information, about this person could be found in databases managed by the Social Security Administration, Homeland Security Department, Internal Revenue Service and perhaps even the FBI.But isn't the problem with data, which prompted post-Sept. 11 criticism, that one database doesn't know what's in the other database?The integration of data within a federal agency and sharing it among various agencies is a long-standing problem. Master data management, or MDM, is one response.'The panacea of MDM is to identify the true single source of the truth for the data and to use that one source as the source for all applications and programs,' said Brad Cole, a solution architect manager at Oracle.And the seemingly simple task of identifying this single source of truth can have profound effects ' if done correctly.'Until recently, most government agencies were more focused on building a data warehouse and reporting against that data warehouse,' said Mike Protz, director of U.S. government support at SAS Institute. 'Now they are moving beyond basic warehousing and reporting and focusing more on the advantages of MDM as an enterprise strategy. With the growth of service-oriented architecture and systems integration, MDM has garnered more attention.'Managing data silos, a subject familiar to government managers, is the end game of MDM. Putting the concept in motion can be overwhelming, however. Steadily, vendors and consultants see it as a new opportunity. Yet others see MDM as hype and today's way of capitalizing on an age-old problem.[IMGCAP(2)] 'MDM, although known by various monikers before the buzzword came into use, has been around for years and is another step in the evolution of 'getting control' of all the stovepipes and legacy-system data that exists in federal agencies' systems' said Tom Kyte, vice president of public sector strategic sales support at Oracle.Dr. Peter Aiken, an associate professor and researcher at Virginia Commonwealth University's information systems department, conducted research about data's role in the enterprise, and said that data problems are nothing new.'There's lots of guidance out there for MDM that is just plain wrong,' Aiken said. 'MDM doesn't solve your problems. Good guiding principles do.'Nevertheless, with Microsoft's recent acquisition of pioneering MDM company Stratature, it seems that many vendors want to get a piece of this emerging niche within government information technology.'Everyone is a player,' said Lance Osborne, marketing leader at CDI Solutions, which provides consulting and implementation services for data integration and MDM. 'Oracle, SAP and IBM are at the top of most analysts' lists. Other big vendors are sniffing at the door, including Microsoft and [Hewlett-Packard]. There are also major players on the systems integrator and consulting side of this trend, including Accenture and BearingPoint.'At the moment, deploying MDM is not synonymous with deploying a specific technology. It begins with a good understanding and definition of who owns the data, and who is to see and use that data. It requires a process-centric solution of defining the 'master' source and then identifying the other systems that rely on that data. Once ownership is set and controlled, technology tools enter the picture to enable the ownership and sharing of data.'The idea is that MDM is the method by which you have clean, accurate consistent data synchronized across an enterprise,' said Bill Cooper, vice president of data warehousing at Teradata.Cooper said he has seen wildly diverging answers to seemingly simple questions, such as the amount of a particular invoice or how many flying hours a military pilot logged. In many cases, you will get different values for the same thing, depending on which system you query. 'If you don't have clean, consistent data across the enterprise, your ability to manage and report on the data is limited,' he said.[IMGCAP(1)] MDM relies on standardization, governance, techniques, tools and a technology platform that addresses the issue of data usage, security and quality, said Bill Kirkendale, chief information officer at the District of Columbia's Court Services and Offender Supervision Agency (CSOSA). Done correctly, it can provide seamless interoperability, data exchange, and transactional and analytical integrity.For the CSOSA office, the importance of MDM came into focus when it embarked on building an enterprise data warehouse to consolidate material about offender supervision and treatment, which was held in multiple databases, Kirkendale said.One of the biggest worries was balancing the privacy of individuals in the system against the distribution of data that would be valuable for maintaining the public safety.Compiling data in an enterprise data warehouse could present a headache for program managers. There are many policies and laws concerning the safeguarding of personnel data that agencies such as CSOSA must adhere to. When used in only one system, the program manager can assure that use of that data meets the appropriate guidelines. Once that data is distributed, though, control gets a lot more difficult.'I need a traffic cop, an arbitrator and an adjudicator to make decisions in real time,' Kirkendale said. 'The decisions must be informed and they must be informed based on today's conditions, not the conditions that were.'Having MDM in place and overseeing the data can simplify this task immeasurably.Aiken said MDM is being sold at the chief information officer level but needs to be embraced at the operating level. Before you throw money at a vendor's technology solution, take a good look at your data architecture. Embracing MDM before you look at your architecture, Aiken said, is like trying to put 'old wine in new bottles.'Aiken likens successful use of MDM to solving a jigsaw puzzle. You begin by gathering the edge pieces, grouping them and understanding the structure that results. Once you understand your data architecture, only then should you sign a contract for MDM software.This task is not simple, however, especially because the data can be shared across other offices within the agency or with other agencies altogether.'Most agencies have done this to a large extent within their own boundaries and are now attacking the challenge of mastering data that is shared among multiple agencies,' Aiken said. Many agencies are investing in the physical tools to share information across boundaries, such as data warehouses. This reuse of data in new circumstances leads to all sorts of new questions about the meaning of the data.'Federal agencies have been putting together their data for quite some time to make it accessible and understandable from many perspectives and/or dimensions,' Kirkendale said. 'In the most recent years ' with the Congress and the administration's focus on accountability and performance ' special attention has been placed on the organizational, programmatic, human resource and financial interpretive power of the data.'Kirkendale said he believes that MDM is a fundamental building block of enterprise architecture, and that getting the architecture right needs to come from the top down.'Though mandated, it is finally sinking in and moving beyond its compliance-purposed adolescence,' Kirkendale said of enterprise architecture.'The dots need to be connected and the language common ' bring the EA and the enterprise data warehouse folks together,' Kirkendale said. 'Give them a side task: Do MDM.'
ON TRACK: The District of Columbia's Bill Kirkendale says master data management helped streamline development of a data warehouse to track criminal offender supervision and treatment.
GCN Photo by Zaid Hamid
'MDM ' is another step in the evolution of 'getting control' of all the stovepipes and legacy-system data that exists in federal agencies' systems' ' Tom Kyte, Oracle
GCN Photo by Olivier Douliery
Practice, practiceTaste the wineJim Romeo is a freelance writer based in Chesapeake, Va.
GCN assistant managing editor Joab Jackson contributed to this article.