You are on page 1of 18


By: Harrison Reid

Y › at is a Data ›are ouse Arc itecture

Y Five Main Data ›are ouse Arc itectures

Y Factors T at Affect C oosing A Data ›are ouse

Arc itecture

Y Summary

Y Bibliograp y

Y Àrimarily based on t e business processes of a

business enterprise

Y Conceptualization of ow t e data ware ouse is

6  ?    

Y ©ndependent Data Marts

Y Data Mart Bus Arc itecture

Y Hub-and-Spoke

Y Centralized Data ›are ouse

Y Federated Arc itecture


Y Data marts t at are independent of eac ot er

Y Often created by organization units

Y ©nconsistent data definitions and different

dimensions and measures
©?    ?  

©ndependent data
End-user access
Source Staging marts
systems area (atomic/summariz
ed data)

Y Creation starts wit a business requirements

analysis for a specific process suc as orders,
deliveries, customer calls, or billing.

Y One mart is created for a single business process

Y Additional marts are developed using t e

conformed dimensions of t e first mart

data marts linked
End-user access
Source Staging by conformed
systems area dimensions
ed data)

Y Developed after an enterprise-level analysis of data


Y Focused on building a scalable and maintainable

Y Developed in an iterative manner
Y Dependent data marts obtain t e data from t e
ware ouse
Y Consist of a centralized ub t at accepts requests from
multiple applications t at are connected t roug spokes
M ?  

End-user access
Source Staging relational
systems area ware ouse
(atomic data)

Dependent data marts

(summarized/some atomic data)
Õ ?    

Y Similar to t e ub-and-spoke arc itecture except

t ere are no dependent data marts

Y Contains atomic-level data, some summarized data,

and logical dimensional view of t e data

Y Queries and applications access data

Õ ?    ?  

relational End-user access
Source Staging
ware ouse and
systems area
(atomic/some applications
summarized data)

Y eaves existing decision-support structures in place

Y S ares information among a number of different


Y Data is eit er logically or p ysically integrated

Y S ared keys
Y Global metadata
Y Distributed queries

Existing data ogical/p ysical

End user access and
ware ouses, data marts, integration of common
and legacy systems data elements
Y ©nformation ©nterdependence between
Organizational Units

Y Upper Management͛s ©nformation Needs

Y Urgency of Need for a Data ›are ouse

Y Nature of End-User Tasks

Y Constraints on Resources
Y Strategic View of t e Data ›are ouse Àrior to

Y Compatibility wit Existing Systems

Y Àerceived Ability of t e ©n-House ©T Staff

Y Tec nical ©ssues

Y Social/Àolitical Factors

Y T e Data Mart Bus, Hub-and-Spoke, and Centralized

Arc itectures are t e most used

Y Many factors affect t e c oice of a Data ›are ouse

Arc itecture

Y Some Data ›are ouse Arc itectures can be

implemented on existing systems