Exploration Data Warehouses
Non-Power Users May Require Separation of Data for Exploration
May 14, 2009
Explorers are 'out-of-the-box' thinkers who may want to submit 72-hour queries. If there are only a few 72-hour requests, or if exploration processing is not a priority item, then these queries can be handled in the enterprise data warehouse.
The Exploration Warehouse
If the 72-hour query becomes a regular feature and if exploration becomes a regular occurrence, then it may be expedient to create an exploration warehouse. An exploration warehouse is easily created from the enterprise data warehouse, assuming the enterprise data warehouse has been created at the proper level of granularity. If the enterprise data warehousing contains very detailed data, then it is a straightforward process for an explorer to create an exploration warehouse by placing a subject of data peculiar to his/her analytical needs into the exploration warehouse.
Creating exploration warehouses can be a regular part of data warehouse life. By creating the exploration warehouse, the data warehouse administrator provides a positive way to handle the workload disparity between high-power explorers and regular users.
The Metadata Infrastructure
Another important activity that is a part of the on-going management of a data warehouse environment as the data warehouse matures, is the creation and population of a metadata infrastructure. In most cases data warehouse administrators are so busy during the initial design and population of the warehouse that little or no attention is paid to the metadata that resides inside the data warehouse. Issues to address when creating a metadata environment include:
- Is there support of both business metadata and technical metadata?
- Will metadata reside at every server in the decision-support environment, at each enterprise data warehouse, and each operational data store?
- Can the metadata at each server communicate and share metadata with every other server?
- Is there a system of record in which any unit of metadata is owned by exactly one server?
Organizational Growth
When a data warehouse grows and matures, so too does the data warehouse administrator function. There are several signs of growth within the data warehouse administrator organization. They include:
- An increase in the number of people doing data warehouse administration work
- A transition from data warehouse administrator generalists to data warehouse administrator specialists
- Increased understanding of the data warehouse administration tasks
The administration requirement of the warehouse environment may grow for several reasons, including:
- Complexity- as the different types of data grow, the complexity of managing them also grows
- User demands- as the number of users grown, their demands for data grow, creating challenging and ever-expanding administrative work
- Age- as data ages, its structure changes and requires administrator work.
Proactive Administration
Meeting the requirements of on-going administration and management of the data warehouse, as described above, are key issues in maintaining the health of a centralized data warehouse. It’s important to broach these considerations as early in the warehouse’s life cycle as possible- optimally at the design stage. Once the warehouse is built and is being administered, it is too late to go back and reengineer the warehouse.
 |
