29-12-2012, 05:05 PM
STUDY OF DIFFERENT TOPOLOGIES OF DATA- WAREHOUSE
STUDY OF DIFFERENT.docx (Size: 42.48 KB / Downloads: 19)
Centralized data warehouse.
• A data warehouse contains properly conditioned data on subjects of interest to users within an enterprise’s multiple business units.
• A data warehouse supports cross functional information requirements.
• A centralized data warehouse can be topologically simple, because it is the sole locus of warehouse data for many users (or clients) and applications throughout the organization.
• It is the topology of choice for many large data warehouses seeking the advantages of economy of scale and centralized system management.
Data warehouses and data marts.
• A data warehouse is often contrasted with a data mart, which typically contains a narrower scope of data, characterized by a single subject, a single business function, or even a single application.
• Data marts are often connected to a centralized warehouse in a three-tier configuration in which clients are connected to specific data marts that draw their data from a data warehouse.
• This topology exploits locality of reference to provide optimal performance to the warehouse’s data-mart clients while allowing user access to warehouse data in order to meet cross-functional information requirements.
• It also facilitates delivery of cleansed and reconciled data to the dependent data marts supported by the warehouse
• . Data marts are managed on the second tier of servers in this configuration and often support online analytical processing (OLAP) and more advanced analytic functions.
Distributed data warehouse.
• This topology consists of network-connected data warehouses with strong distributed processing support.
• In its more advanced forms, it enables users at any location or clients connected to any data warehouse to work as if the data resided in a single, centralized enterprise warehouse, although it is physically distributed among multiple data warehouses. Such distributed, or virtual, data warehouses
require strong capabilities for distributed database management; at its best.
• This topology should provide users a single-site image of the global warehouse. When weighing deployment of a distributed configuration,it is important for implementers to take into account the internode database processing that will be needed to meet user requirements.
• Performance requirements can easily preclude adoption of this topology for applications requiring frequent distributed joins and other distributed set operations.
• It is feasible only when designers are convinced that remote distributed set operations are required infrequently and for light data loads.
• When the warehouse data stores are managed by database management systems (DBMSs) from multiple vendors, the distributed data warehouse topology benefits from use of a multidatabase server (MDBS) that simplifies user access to heterogeneous data sources.