Database Reference
In-Depth Information
data. This allows you to leave intact any data that does not require
cleansing, which may help to conserve processing resources.
Disparate data domains : In cases where data is structurally similar
but syntactically different, you might want to employ branching to
handle the data differently within your data flow. Consider the ex-
ample of geographical address data: although they both describe a
physical address, you might need to process domestic addresses dif-
ferently than you would handle international addresses. By using
branching tools such as the conditional split, various address types
from a single source type can be handled within one data flow task.
Varying metadata : Although relatively rare, there will be the occa-
sion where a source may contain rows with varying metadata. Con-
sider a text file with a ragged structure in which some rows are miss-
ing columns at the end of the line. By splitting the data based on the
absence of certain columns, you can account for the metadata differ-
ences inline.
Figure 10-6 exposes this design pattern by showing the use of expression logic to
break apart a data stream into multiple outputs. In this case, you are processing a
billing file by using comparison expressions within the conditional split transformation
(see the callout) to determine whether each row is paid on time, not yet due, or past
due, and then you're sending it to the appropriate output accordingly.
Figure 10-6 . Using expressions to define multiple paths
 
 
Search WWH ::




Custom Search