In Ai, metadata is an important means and basis for the smooth completion of data integration between various business systems. It is the key to ensuring data quality. Effective metadata management can smoothly reflect the changing demand into the data warehouse. Come. In a data warehouse project, guarantee the unity, effectiveness and normative management of metadata is the key to the success of the entire project.
Then, how do the metadata in the data warehouse process in different stages do effectively management, what are their management methods and ways? What is the difference between the existence form or management of the existence or management of the existence or management of different stages (such as demand analysis phase, model establishment phase, ETL phase, data mining and front-end display)?
The management of metadata should be the focus and core of the whole process in the data warehouse; always feel the actual project, although the designer will point out the importance of this link, but the real development and implementors seem to be One core link has not yet formed a clear understanding, or there is no scientific approach and tool to manage, just record metadata in the form of some files, may this be a performance of the data warehouse in China in the primary stage?
The ETL tool is your own dollar data management tool or a set of ways. The repository provided in some tools is the management of metadata in the ETL process; but this part in a data warehouse project is only a stage of Yuanyuan. Data management, and only apply to this phase, the fact situation is that metadata is often applied to every stage of the entire project, so we think about how to concentrate the metadata in each stage during the process and Effective management? If a good tool or solution, it can easily understand and accept metadata, so that the user accepts the reliability of data quality in the data warehouse, it will respond more quickly to the demand for users, and it also facilitates the overall maintenance of the project. .
The above article has a relatively clear understanding of the DW a few months ago, and actually participated in the thinking of DW, the original is posted to DwWay and CNOUG, I want to throw the jade, let Many buddies with DW projects talk about their own feelings and opinions, and they have little response, and they hope that
Below is a friend who participated in the discussion proposed some views and opinions, and it feels quite quality, and it is also excerpt from:
Socket organizes the use or function of the metadata in the data warehouse field:
1) Describe which data is stored in the data warehouse
2) Define data extraction and conversion to involve the management of mapping between the operating environment and the data warehouse environment.
3) Describe data synchronization requirements
4) Record changes in the data structure in the data warehouse.
5) Measure data quality
6) Data warehouse enables registration, access, and management of external data through metadata.
6) Another data type associated with metadata is "notification" data
