Effective Metadata Management has increasingly come to include enterprise-wide integration. A report from Gartner states, “To create greater value, you must move from managing metadata in silos to a more integrated, enterprise wide approach.”
Providing a point of commonality in the terms, definitions and rules that an organization’s metadata is based on reduces redundancy, incorporates greater sources and data types, and aids agility.
Doing so correctly requires finding a delicate balance between utilizing a circumscribed Metadata Management approach (in which respective departments and organizational functions have complete autonomy in creating definitions and rules) and an austere one in which relevant metadata is stored and accessed via an enterprise-wide repository.
Although there are obvious shortcomings with both of these extremes (the former defies governance and sacrifices quality, the latter is extremely costly and time consuming), a hybrid approach that evolves over time from one to the other provides an ideal medium.
How each organization engages in this process depends on its needs and resources, both technological and otherwise. Key factors that determine the most useful strategy involve prioritizing the business, mitigating governance concerns, and selecting relevant metadata.
Business First
Like most every other facet of Data Management, the objective of Metadata Management is to serve the ends of the business. Therefore, the pace at which an organization evolves from a siloed metadata approach to an integrated one is largely determined by the effects on the business related to changes (and possible delays) on processes, projects that need to be completed, ROI, business sponsorship and user importance.
The three principle types of metadata include that pertaining to operations, technologies, and business objectives—which may include specific rules, repositories and definitions that assist this part of the organization. It is essential to prioritize business metadata before the other two, and to slant technical and operations metadata in such a way that they are in congruence with (and can ideally aid) business metadata and processes.
Additional prioritization factors for metadata management that will assist business professionals include:
- Upper level management and business support: effective metadata management requires more than just support from a CIO, especially when attempting enterprise-wide integration. Such initiatives benefit from the auspices of upper level management as well as the business—which is why it is key to outline and target ROI and early, tangible objectives that can justify potential spending on technology and restructuring. Assistance from IT is also invaluable in facilitating integration.
- Identifying specific business objectives: It is critical to base what metadata is shared throughout the enterprise around specific business objectives, which may be either project-based, long term, or both. Develop a set of metrics that will help business professionals achieve these aims, and ascertain what data throughout the rest of the enterprise is relevant to their reaching these goals. Doing so will clarify how long integration should take.
- Assess the immediate and long term impact of changes on the business: integrating metadata throughout the enterprise is bound to affect processes and procedures. Determine the outcome of such changes on the business, and structure them accordingly so that they do not negatively impact objectives. A general level of hierarchy when considering changes is to prioritize the business, IT, and then operations.
Judicious Selection
With the amount of data produced growing almost exponentially, it is highly impractical to attempt to integrate all of an organization’s metadata. According to Mike Blechar of Gartner:
“most…metadata needing to be managed exists in a machine-readable format in disparate technologies having inconsistent metadata standards and limited ways to share/integrate their metadata. Complicating this further, most organizations have inconsistent siloed terminology/taxonomies/ontologies/semantics across business units, products lines and application systems and data.
The focus on Metadata Management, therefore, should be on sharing only that metadata that are relevant throughout the entire organization–and which serve specific business purposes. Examples include metadata that are redundant throughout various departments as well as that which are used in various organizational processes. Ideally, integrated metadata functions as a subset of metadata pertaining to Enterprise Information Management.
Integration Approaches
The following is a list of methods for integrating metadata ordered in a gradual progress of time commitment, technological and human resources, and sophistication which will take an organization from a siloed approach to the quintessential central repository. The initial methods can considerably help in departmental metadata environments, which is why most enterprises should start with these before progressing to the later ones.
- Implement common definitions and naming conventions within siloes. This approach provides a point of commonality throughout the enterprise and can be done at the pace of the respective departmental users. Although the metadata aren’t necessarily integrated, they can become substantially more uniform so that when it is necessary for say, business analysts to access a MDM tool, the common terms and definitions will aid them considerably.
- The most rudimentary aspect of metadata integration is to implement a web browser that can offer views across different tools (such as data warehouses). Although the metadata still aren’t fully integrated with this approach, this approach facilitates sharing.
- Utilize bridges between various tools (especially those related to Business Intelligence and warehousing) to share metadata. This is a more advanced approach because users are often dealing with tools from various vendors that use different technologies, which explains the greater time commitment and involvement of technological and IT resources. However, this is the first step towards true integration, and is more cost efficient than the subsequent ones.
- Federating multiple repositories is a highly effective means of integrating metadata. This option involves purchasing respective repositories from metadata vendors by organizational domain, and inputting enterprise-wide and business-focused data into them so that separate facets of an organization have access to relevant data throughout the organization. This method provides definite integration, although there is still some partisanship involved as various enterprise domains have their own repositories.
- Lastly, it is possible to consolidate all relevant metadata (essentially that which is housed in various domain repositories) into a single repository to offer cross organizational access and view to enterprise-wide metadata. This is the ultimate option for metadata integration and typically takes years and considerable funds to implement. However, organizations can best manage their metadata so that they can accomplish this option by enacting the measures discussed in the previous options.
Governance Concerns
It is impossible to have integrated metadata management (or even effective Metadata Management) in a siloed approach without efficient governance throughout the entire enterprise. In this respect, the seeking of successful Metadata Management is an integral part of meaningful Data Governance. One of the benefits of siloed Metadata Management is that there are clear boundaries regarding the ownership of data—which supports the first couple of integration approaches outlined above. Governance complications related to integrated metadata revolve around making changes and how those changes impact disparate users and their data that is dependent on an altered data element.
Governance measures must be taken, therefore, to authorize when metadata changes can be made, in what form, and how they are communicated throughout the organization. These concerns particularly apply to metadata that are deemed enterprise-focused and apply less to that which is relevant to specific domains. When selecting which metadata to integrate that which already has firm governance in place supersedes that which does not.
Retrospect
Successful metadata management requires integration, or at least an alignment of domain metadata via common rules, definitions and terms. When seeking to achieve these aims it is necessary to prioritize business concerns, and base the timing and the amount of resources dedicated to this process on specific outcomes and goals of the business. Aligning metadata in a uniform way is more of an evolution than a project; metadata integration should only involve the most salient data that effects business processes. Firm governance measures should be in place to provide common definitions and procedures, as well as rules regarding changes to metadata.
