The steps in the MDM process are as follows

Land

Download the input table files required by the organization from an external source to the Informatica MDM Centre. This is basically collecting data from external sources for processing. This phase is called the grounding phase because the data is collected into a table in the node repository called the grounding table. This process is performed by a separate ETL tool.

Stage

In this case, the data is read from the grounding table and cleaned to make it meaningful and specific. The cleaned data is then read into the corresponding table, called a one-step table. When data detection is started or changed, the CDP is run and the data is written, then only the changed CDP is run and a new write is done, the unchanged write remains.

Load

In this step, the data from the configuration table is loaded into the corresponding table in the central memory (usually called the base or dependent object). Trust rules are used to determine trustworthy data.

Tokenise

A set of matching keys results in a token that is used by the matching process to identify the base objects to be matched.

Match

In this step, the records are compared against the matching rules. Duplicate records can be found with the help of separate flags.

Consolidate

A consolidated record is created by combining the values of duplicate records. This consolidated record contains most of the valid cell values from the original record or from the record in the level table.

Publish

Publish BAT to other systems as required. For batch processing in Informatica MDM, this is implemented as a stored database that can be accessed from the Hub console or from another user script.

Real-Time Processing

For real-time processing, external applications communicate with the Informatica MDM Hub through an interface called the Service Integration Framework (SIF). The SIF has several APIs that interact with the MDM Hub to read, clean, append, match, update data, and so on. Real-time processing can be used whenever a record is updated from a data source or a record is deleted from a source system. Therefore, real-time incremental processing in MDM can be used to process incremental workloads in the central warehouse. Incremental loading is nothing more than processing newly added, updated, or deleted data from a data source.