Integration of Heterogeneous Databases in Data Warehousing
We have two approaches to integrate two different heterogeneous databases.
- Query-driven Approach
- Update-driven Approach
Query-Driven Approach in Data Warehousing
Query-Driven Approach is the traditional approach to integrate heterogeneous databases. The Query-Driven Approach approach was used to build mediators on top of multiple heterogeneous databases. These mediators are also known as integrators.
Process of Query-Driven Approach
Process of Query-Driven Approach involves the following steps’
- To make a query to the client-side
- Translation of the query by the metadata dictionary. Metadata dictionary translates the query into a suitable form for different heterogeneous sites(that are involved in the integration process).
- Mapping of the queries are mapped
- Mapped queries must be sent to the local query processor.
- Collection of the results from different heterogeneous sites.
- Integration of all collected results into a global answer set.
Disadvantages of Query-Driven Approach in Data Warehousing
- The query-driven approach needs complex processes. Two common common complex processes are integration and filtering processes.
- If your quires are frequent, then a Query-driven approach is very expensive.
- The query-driven approach is also very expensive for queries that require aggregations.
- The query-driven approach is very inefficient.
Update Driven Approach in Data Warehousing
Update-Driven Approach is an alternative to the traditional Query-Driven Approach. Nowadays, the latest data warehouse systems follow the update-driven approach. In an update-driven approach, the information from many heterogeneous sources is integrated in advance and then the information stored in a warehouse. This information is available and very helpful if we want to query and analyze the information directly.
Advantages of Update Driven Approach
This approach has the following advantages −
- Update-Driven Approach provides high performance.
- The following steps are must for data.
- Copying the data
- Processing the data
- Integrating the data
- Annotating the data
- Summarizing the data
- Restructuring the data.