In the domain of information technology (IT), metadata is about data and other IT artifacts. The most common definition of metadata is “data about data.” However, because data without context is merely a meaningless sequence of symbols, the discussions about metadata quickly revert to “information about data, information about information.” If you have a large collection of information organized in a way that you can extract relationship and data flows and perform explorations and investigation so that it becomes actionable information, you can refer metadata as “knowledge about data.”
In the domain of information processing, the objects are electronic records of data assets. Such data assets include data files, data columns in a database, business intelligence (BI) reports, or a transformation job that pours the content of one data store into another. These data assets are the type that an enterprise creates, maintains, and uses to conduct business. The information about these data assets is most valuable to separate users within the organization. Data analysts and developers want to know the physical characteristics of these assets.
Business metadata includes business terms and their definitions, examples of usage, business rules policies, and constraints. Altogether they define the semantics of a business concept and its realization in physical data assets.
Technical metadata consists of the technical description of data assets. Technical metadata includes the following descriptions:
- Schemas, tables, and file layouts
- Source and target datastore identification and physical attributes
- Data mappings
- Formal specifications of transformation jobs, business rules, and other
Operational metadata consists of information about the execution of an application or a job. Such information includes times and dates, counts of records processed and rejected, and statistics about processes, processors, and servers that are generated in the course of executing a job.
Operational metadata is used to answer the following questions:
- At what date and time was a job last executed?
- How many records were processed?
- How many records were rejected?
- How long did it take to process the job?
Source : IBM Redbook, Metadata Management with IBM Infosphere Information Server http://ibm.com/redbooks