Let’s put information back in IT

IT means Information Technology, but often IT departments focus on the technology side and don’t give the information side equal attention. The information side deals with the form and format of data and with the data’s “metadata.” Metadata is data that is used to explain and describe other data.

By Dennis Brandl September 1, 2007

IT means Information Technology, but often IT departments focus on the technology side and don’t give the information side equal attention. The information side deals with the form and format of data and with the data’s “metadata.” Metadata is data that is used to explain and describe other data. For example, one piece of data may be an employee ID called “EmployeeID” that can have a specific value, such as 1999000023. The metadata for EmployeeID defines the type of data (a 10 character numeric string) and any rules the data must follow (the EmployeeID must start with the 4 digit year of first employment, and no two EmployeeIDs may be the same). IT groups manage information by coordinating and controlling values of the data and ensuring that all data uses conform to metadata rules. Coordinating and controlling this information across several different applications can be difficult and is often the hardest part of application integration. In the manufacturing IT domain, one common difficulty is managing equipment tag names because every application seems to have its own rules and restrictions and these can often be in direct conflict.

Shared information is usually separated into two categories: master data and transient data. Master data has several characteristics: It describes business objects that are important to multiple business processes; key parts of the business object must be shared across multiple applications; the complete business object is not completely described by a single application (different applications maintain different elements); the business object can be uniquely identified; and the data is relatively stable, often with a lifetime measured in years. In manufacturing systems, master data often includes material master definitions, physical asset identification, equipment classes, people, and personnel roles.

Other shared data is called transient or transactional data. It changes relatively often and can have a limited lifetime. The line between master data and transient data is fuzzy, sometimes leading to confusion in deciding what data is master data. For example, a physical asset definition could be master data, while the maintenance status of the equipment could be transient data.

IT departments have developed a method for coordinating and controlling master data called master data management (MDM). Unfortunately, the MDM acronym has become an often misused and misunderstood marketing term. MDM is not a technology. It is a collection of management processes that can be supported by data management tools. Don’t be swayed by MDM vendors into believing that a single product will solve your data management problems. Without well-defined management processes, all an MDM tool will do is cause confusion and conflict.

Management processes must be defined to handle adding new master data elements, deciding when to remove master data, enforcing rules for the use of master data, ensuring that no alternate uses and definitions creep into applications, checking all shared data to see what should become master data, and checking all shared data to see if it already is defined as master data (usually with a different name). Companies that have successfully implemented MDM have set up dedicated MDM process teams. The teams are given sufficient authority to require all projects to use master data correctly. MDM process teams do not look at any specific technology, because other groups are responsible for specific MDM tools.

MDM is a critical element for effective application integration, especially for manufacturing to business system integration. If your company has an MDM team, make sure it understands manufacturing’s master data requirements and that manufacturing master data processes are included as part of overall MDM processes.

Author Information

Dennis Brandl is president of BR&L Consulting in Cary, NC, dbrandl@brlconsulting.com .