276 Solving Operational Business Intelligence with InfoSphere Warehouse Advanced Edition
7.3.1 SQW data flow
Figure 7-1 on page 275 shows a simple data flow that extracts sales transactions
from a source database and loads that data into a target data warehouse fact
table.
???? A Table Source object is used and configured to select data from a source
table ORDER_HEADER.
???? A second Table Source object is used and configured to select data from a
source table ORDER_DETAILS.
???? A Table Join object is used and configured to join the ORDER_HEADER and
ORDER_DETAILS data flows and provide a single data flow for each
transaction within an order that is aligned with the target fact table.
???? A third Table Source object is used and configured to select datetime data
from the date dimension.
???? A Fact Key Replace object is used and configured to replace the date in the
data flow with a date key to join the target fact table to the date dimension.
???? A Bulk Load object is used and configured to load the transformed data into
the staging table.
The data can then be inserted into the production tables from the staging tables.
In this way, the production table does not have to be taken offline, full query
access is supported, and recoverability is maintained because the INSERT
statement is a logged transaction.
This simple data flow represents a discrete component of what is a larger
application; extracting, loading, transforming and inserting data. By developing
further discrete components to process transformation and loading of dimension
and additional fact data, you are positioned to perform these tasks:
???? Change, debug, and monitor at a component level.
Having several smaller ETL components rather than a single larger
component allows for changes to be isolated and debugged at a more
granular level.
???? Manage and control parallelism at a component level.
Where significant volumes of data are presented for only one of several data
sources, for example, you can increase the parallelism at which data is
processed for just a single data source.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.