This article originally appeared on the BeyeNETWORK.
New business performance management (BPM) software solutions allow financial reporting applications to track an extensive amount of dimensional data. Finance’s need for detailed analyses can cause these applications to track data that overlaps with other applications, data marts and warehouses. This can complicate an organization-wide effort to create a viable managed metadata environment (MME). The dilemma many IT areas face is how to satisfy finance’s requests while preserving organization-wide data integrity and keeping an MME effort on track. Fortunately, tools that focus exclusively on master data management can enable a more democratic approach, increase end-user satisfaction in finance and establish a more participative MME, while simultaneously helping to reduce IT’s ETL (extract, transform and load) burden.
IT’s Toughest Metadata Customer
Sarbanes-Oxley Act (SOX) initiatives have given the finance area good reason to be demanding; but from an IT perspective, SOX compliance requests can result in yet another large data mart or marts that need to be managed, maintained and integrated with the organization’s current or planned MME. These larger data marts and other possible BPM applications are necessary to allow finance to track greater detail, further assuring financial statement accuracy. SOX compliance also reinvigorates finance’s need for metadata management because first phase 404 compliance required process controls that necessitated the documentation of certain workflows. However, any effort to strengthen these controls typically highlights the need for an MME for automation of these controls, making them sustainable and predictable. All this translates into a great deal of joint IT/finance effort, but this work doesn’t need to overload IT’s already extensive workload.
IT necessarily addresses metadata management needs with data warehousing (DW) best practices; however, DW best practices typically fail to describe a universal solution for metadata capture and management. (Source: R. Kimball and J. Caserta, The Data Warehouse ETL Toolkit. Indianapolis IN: Wiley, 2004.) They do, however, stress the key role metadata plays in any DW or MME endeavor and the difficulty in determining who is primarily responsible for business metadata. Any discussion regarding how many data marts are necessary, what data they contain and how the data in each of these applications relates the other data inevitably requires critical decisions to be made regarding metadata management. Even though commonly accepted DW practice recommends that metadata owners agree on a single definition of metadata from an enterprise standpoint, it’s easy to see why this is extremely difficult to accomplish. Different metadata definitions for the same data element will inevitably, and sometimes necessarily, exist – the issue is to manage this necessary metadata diversity while simultaneously centralizing it.
Centralized versus Managed
IT has traditionally been burdened with enforcing metadata standards when their users in finance and other areas require the flexibility to develop their own analysis applications. Centralized control of this master metadata will place limits on the efficacy of the applications in finance and elsewhere; no control of metadata can lead to data duplication and overall data integrity problems. Instead, a “virtual centralization” is necessary, but traditional DW approaches make centralized metadata definitions difficult to establish. Furthermore, IT often must assume extensive ETL efforts to maintain them. The establishment of a workable MME is possible if a master metadata management approach is adopted that will allow each functional area, including finance, to manage its own proprietary metadata definitions.
Maintaining the Mapping
IT needs to enforce mapping standards by playing a lead role in establishing master data definitions, but they certainly can’t map finance’s (or any other department’s) source system metadata on an ongoing basis. The business user is most familiar with business metadata, so it follows that only the business user can do this in a consistent, accurate manner. This is problematic, however, since most commonly used ETL tools are not intended for functional end users.
Central to the concept of master data management (MDM) is the availability of end-user friendly software that takes this burden off IT’s shoulders and assigns it where it belongs – in the hands of the business users. IT will need to establish and monitor the organization-wide adherence to master data management standards, but it will not maintain the mapping.
Where Does ETL End and MDM Begin?
Although some functional overlap will exist, ETL and MDM have fundamentally different purposes. ETL software allows for a high degree of data cleansing and for multiple applications to share data; MDM software allows for the creation of an “application superset” in which cleansed master metadata is defined, related to the metadata in participating applications and maintained. A fundamental difference with the new MDM tools is that they are also intended for use by the business user. The MDM software serves the purpose of automating the master data management workflow by monitoring the metadata in participating applications and notifying the pre-assigned business users to any changes. Then, using a graphical user interface (GUI), users can update their participating application’s metadata to accommodate organization-wide master data definitions. Customized workflows, approval procedures and security can be incorporated.
Where to Start
The design of any MME venture begins with an assessment of best practices, current metadata management status and an evaluation of available tools to support the targeted MME. A well established MME can create tremendous value for an organization from a cost reduction, process automation and, of course, an analysis empowerment standpoint. It can also facilitate shared learning, leading to increased organization-wide efficiencies. When mapping out your MME strategy, finance’s swollen BPM applications are a great place to start.