Multi Source Loads With OBIA
Multi Source loads refer to the situation where we load data from two of more source instances of similar or dissimilar types. Some common examples include:
Multi Source loads refer to the situation where we load data from two of more source instances of similar or dissimilar types. Some common examples include:
Ø Loading data from two similar type instances separated geographically for example EBS 11i in US and EBS 12i in Japan
Ø Multi source load from dissimilar instance such as an instance of Siebel 7.8 in one location and of Oracle EBS 12i in another location.
Anyways, there are two key things to be noted in the case of a multi source load and this has primarily to do with the locking mechanism if they are seeking same targets and the truncate table, drop index and analyze table sequence. These details are as follows:
Multi-Source Order of Execution
The order in which DAC loads data from the different sources that are participating in
the ETL process is determined by the priority of the physical data source connection (set in the Physical Data Sources tab of the Setup view). The Priority property ensures that tasks attempting to write to the same target table will not be in conflict.
Note this property is primarily responsible if the sources are of the same types for example EBS11i and EBS 12i then one of them can be given a priority 1 and the next physical data source a different priority
Delay
As shown in the following Figure , the Delay property is located in the Parameters subtab in the Execution Plans tab of the Execute view. You set this property to specify how many minutes an extract of a data source will be delayed after the first extract of a multiple source extract process started.
Folder Level Priorities
As shown in the following Figure, you need to set the folder level priority in the Task Physical Folders dialog box, which you by selecting Tools, then Seed Data, and then Task Physical Folders.
I know that the first question which must be immediately bothering you is if we can set the priority at the data source connection level itself then why do we need the priority at the physical folder level, well indeed there is a very specific business case for that. Imagine that you have a Seibel 7.8 and an EBS 12 implementation in US plus an EBS 12i implementation in Japan. Then what should be the best mechanism to set the priorities that these loads run in a mixed mode (combination of parallel and serial mode) and the performance is optimum. What you might have to do is first set the priorities at the data source level for the EBS sources for Japan and US, as it’s the same physical folder there is no way that we can set different priorities at the physical folder level, next you need to set the priority for Seibel at the physical folder level.
