freshidea - Fotolia
Align enterprise data architecture, governance for 'quick wins'
Data management consultant Donna Burbank outlines how effective data governance hinges on the deployment of a comprehensive enterprise data architecture.
You can't govern data well if you don't have a good enterprise data architecture. Conversely, successful data governance can be the glue that holds together a data architecture and the data management environment it underpins.
In organizations, data governance is often seen as focusing on one of two things: committee meetings and data stewardship roles, or technical data management and controls, according to Donna Burbank, managing director of Global Data Strategy Ltd., an information management consulting company based in Boulder, Colo. Burbank said holistic data governance combines both of these aspects, and a hearty data architecture and the accompanying diagrams help bind business- and IT-centric governance together.
Indeed, as she sees it, a high-level enterprise data architecture provides the roadmap for data strategy and associated governance initiatives and acts as a guide for small, targeted projects in which data management teams can demonstrate the business value of those efforts. And effective data governance becomes a bigger priority as a data architecture grows.
"As a general rule, the more data [that] is shared across and beyond the organization, the more formal governance there needs to be," Burbank said during a session that was part of Dataversity's inaugural Data Architecture Online event.
Currently, modern organizations use a wide variety of data platforms -- notably, on-premises relational databases, spreadsheets and packaged applications, like ERP and CRM -- according to survey data from a report coauthored by Burbank that Dataversity plans to publish in September 2019. Burbank predicted that the variety of data platforms will continue to grow. For example, survey respondents said cloud-based relational databases -- now the fourth most used platforms -- will move to the top of the list in the future.
Governance customization and collaboration
Burbank pointed out that data governance approaches are not the same for all platforms or the data they contain. Each type of data should have its own type of governance model and rules on data sharing and usage, she advised. "Don't overgovern what you shouldn't," she said. "But don't undergovern your master data so no one knows how many customers you have."
A survey conducted for a 2017 Dataversity report, also cowritten by Burbank, found that regulatory compliance and data governance were second on the list of the main drivers for implementing a data architecture, behind only reporting and BI applications.
Burbank emphasized that building and maintaining a successful enterprise data architecture aren't just the job of the data architect; other data management positions, such as data modelers, data governance managers and database administrators, should also play a role in the process.
"Data governance is about collaboration. You need a team of people to raise a data architecture," Burbank said. "It really isn't just about the data architect. We need all those different roles with different perspectives."
Get 'quick wins' on data architecture, governance
In order to better align data with business priorities and improve efficiency, Burbank said companies need to build a balanced architectural design around the key areas of their business, avoiding both an overly academic focus that isn't relevant to business units and a "Wild West approach" that results in data chaos across an organization.
Donna BurbankManaging director, Global Data Strategy Ltd.
She then detailed a variety of models, diagrams and other tools that data management teams can use to document business processes and data workflows. Doing so in targeted projects can help produce "quick wins" that can then be used to sell other parts of a company on the merits of combined enterprise data architecture and data governance efforts, according to Burbank.
These tools include the business motivation model, which identifies an organization's goals, objectives and corporate vision, and the business capability model, which outlines the core functional areas of the organization.
Conceptual and logical data models create an active inventory of data assets and also "tell a story" about the data, making them a useful tool for showing business users what different data elements mean and ensuring that data definitions are consistent in different systems, Burbank said. Customer journey maps and other process models identify key data dependencies in core business processes to "really put data in a business context for the organization," she added.
Mapping data to business processes
Data can then be mapped to key business processes to understand how it's created and used. For example, a CRUD matrix -- short for create, read, update and delete -- shows data at different stages of usage across the organization, which Burbank said can help improve data governance and data quality.
Burbank gave the example of an international restaurant chain that, having realized it had little control or visibility over its menu data, created a central hub for the data and established a new workflow and data governance policies for improved data management. The chain also used business process diagrams to identify the flow of information and CRUD matrices to understand data usage, stewardship and ownership.
Similarly, England's Environment Agency worked with Global Data Strategy to develop data models and data standards to support the consistent publication of open data on key environmental issues. This helped create a common data language across the organization, with benefits that include cost savings, improved data quality and consistency, and increased collaboration between different teams, Burbank said.
In closing, Burbank emphasized that, more broadly, building on small successes early on is key to developing and ensuring long-term sustainability of enterprise data architecture and data governance efforts.
6 best practices on data governance for big data environments