Wednesday, March 13, 2013

Pipeline Partitioning in Informatica

Pipeline Partitioning a mapping in Informatica actually means to Partition the pipeline inside the mapping. Hope you all are aware of the various Partitions available in Oracle, if not please check about Partitions in ORACLE.

A pipeline will contain the source, transformations and targets. Mapping will be subdivided into many pipelines, these are called stage pipeline .After enabling partitioning in stage pipeline when we run the Integration service it runs pretty faster and we can expect more performance. The Integration service will run the partition threads concurrently.

Now it’s about how to enable partition and where to enable partition
  1.     Set the partition points in the mapping
  2.     Set the number of partitions
  3.     Set the partition types
Partition points marks the boundaries that divide the pipeline into stages. By default Integration Service will keep partition at various transformations. Partition points mark the points in the pipeline where the Integration Service can redistribute data across partitions.

When we give Partition points it actually creates various stage pipelines inside the mapping. Now we need to decide how many partitions we need to have inside one stage pipeline. When we add more partitions it increases the performance of Informatica. You can define up to 64 partitions at any partition point in a pipeline.Also note that if we define two Partitions at any partition point, then the remaining partition points will also have two partitions.

By default Integration service will create a default Partition type for each partition point. Also we have the option of explicitly giving the Partitioning Type. The type of partition decides how Integration service will distribute the data across components. Below are the various Partitions available:
  1.     Database partitioning
  2.     Hash user keys
  3.     Hash auto-keys
  4.     Key range
  5.     Pass-through
  6.     Round-robin

No comments:

Post a Comment