Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


Figure 6.15 depicts a world in which there may be an infinite amount of data in the data warehouse. This is appealing because a data warehouse will contain much data (but not an infinite amount!).


There are, however, some considerations. As the data warehouse starts to expand beyond a few processors (i.e., servers), an excessive amount of traffic starts to appear on the network. The increased traffic appears when a request for data overlaps from one processor to another. For example, suppose one processor holds data for the year 1998, another processor for 1999, another for


2000,    and yet another for 2001. When a request is made for data from 1998 to


2001,    the result set for that query must pass over the boundaries of the processor that originally held the data. In this case, data from four processors must be gathered. In the process, data passes across the network and increases the traffic.


day 1




-I l~»~l IT


~    A




A single server holds all the data for the warehouse.





Now two servers hold the data for the warehouse.



I—I I’l


rr



day 2



I—I II



day 3



П


I—1 l-l




1 I—I II




I—I I’l




1 I—I I’l


/



/’    The number of servers that hold the


warehouse data can be expanded ad infinitum (at least in theory) for as much data as desired in the data warehouse.

Скачать в pdf «Building the Data Warehouse»