An InfoSource
combines quantity of information that logically belongs together,
summarized into a single unit. It prepares consolidated data for
updating to the data targets. InfoSources contain either transaction data or master data (attributes, texts and hierarchies).
When you want to load the related data from different datasources to the data targets, Infosource comes into picture.
Role of InfoSource in SAP BW- 
There is a difference in use of InfoSource between BW and BI.
In BW, a DataSource
is assigned to an InfoSource. If fields that logically belong together
exist in various source systems, they can be grouped together into a
single InfoSource in BW, in which multiple DataSources can be assigned
to an InfoSource.
In the BW Processing
Transfer Rules, individual DataSource fields are assigned to the
corresponding InfoObject of the InfoSource. Here you can also determine
how the data for a DataSource can actually be transferred to the
InfoSource. The uploaded data is transformed using transfer rules. An
extensive library of transformation functions that contain business
logic can be used here to perform data cleansing and to make the data
analyzable.
The transfer
structure is used to transfer data to the BW system. The data is
transferred 1:1 from the transfer structure of the source system into
the BW transfer structure.
If you are dealing
with an InfoSource with flexible updating, then the data is updated from
the communication structure into the InfoCube into other data targets
with the aid of the Update Rules. InfoSources with direct updating permit master data to be written directly (without update rules) into the master data tables.
Role of infosource in SAP BI-
In contrast to 3.x InfoSources, as of
Release SAP NetWeaver BI 7.0, an InfoSource behaves like an InfoSource
with flexible update
The data in an InfoSource is updated to an InfoProvider using a transformation.
You can define the InfoObjects of the
InfoSource as keys. These keys are used to aggregate the data records
during the transformation.
1. Data Flow Without an InfoSource:
The DataSource is connected directly to the target by means of a transformation.
Since there is only one transformation, performance is better.
However,
if you want to connect multiple DataSources with the same structure to
the target, this can result in additional maintenance effort for the
transformation, since you need to create a similar transformation for
each DataSource.
You can
avoid this if the DataSource is the same, it just appears in different
source systems. In this case, you can use source system mapping when you
transport to the target system so that only one transformation has to
be maintained in the test system. The same transformation is created
automatically for each source system in the production system.
2. Data Flow with One InfoSource
The
DataSource is connected to the target by means of an InfoSource. There
is one transformation between the DataSource and the InfoSource and one
transformation between the InfoSource and the target.
We
recommend that you use an InfoSource if you want to connect a number of
different DataSources to a target and the different DataSources have the
same business rules. In the transformation, you can align the format of
the data in DataSource with the format of the data in the InfoSource.
The required business rules are applied in the subsequent transformation
between the InfoSource and the target. You can make any changes to
these rules centrally in this one transformation, as required.
3. Data Flow with Two InfoSources
We
recommend that you use this type of data flow if your data flow not only
contains two different sources, but the data is to be written to
multiple identical (or almost identical) targets. The required business
rules are executed in the central transformation so that you only have
to modify the one transformation in order to change the business rules.
You can connect sources and targets that are independent of this
transformation.
