data warehouse architectures

requirements

separation

• analitycal and transactional processing should be take apart

scalability

• the DWH need to be capable of handle large amounts of data

extensibility

• the DWH should be able to host new applications without the need to redesign the hole system

security

• the access to the DWH should be monitored

administrability

• the DWH should be easy to manage

single layer architecture

• the goal is to minimize the data collected in the DWH by reducing the separation between source layer and analysis layer
• to achieve this, a middleware software need to abstract the complexity and the divergency of the Source layer data

PROS	CONS
the space occupation is minimized	no separation between source and analysis layers

two layers architecture

• in this architecture data are extracted from source layers trough ETL and are inserted in a data warehouse layer where they are stored and accessed by the analysis layer
• the source and the analysis layer are separated but there is more space occupied

PROS	CONS
source and analysis data are separated	less space optimization

three layers architecture

• the reconciled layer is added to the two layers architecture that creates a consistent model of the data and separate data extraction problem and integration problem

PROS	CONS
more logical separation between the problem’s management	more data redundancy