Relation

• Visualization graph consists of
  • nodes, we call them endpoints
  • edges, we call them relations
• relation means, that by executing SQL code, data is
  • copied or transformed from source endpoint to target endpoint, or:
  • target endpoint is influenced by source endpoint, but the value is not directly derived from source

Relation types: direct, semidirect and indirect

• See also Glossary for more info

Direct lineage example

Consider this SQL command:

INSERT INTO TARGET_TABLE SELECT COLUMN1, COLUMN2 FROM SOURCE_TABLE

• There are two direct relations:
  • FROM COLUMN1 of SOURCE_TABLE to first column of TARGET_TABLE
  • FROM COLUMN2 of SOURCE_TABLE to second column of TARGET_TABLE

Semidirect lineage example

INSERT INTO TARGET_TABLE SELECT CASE WHEN A = B AND C = 1 THEN D END FROM SOURCE_TABLE

• There are three semidirect relations:

  • FROM A of SOURCE_TABLE to first column of TARGET_TABLE
  • FROM B of SOURCE_TABLE to first column of TARGET_TABLE
  • FROM C of SOURCE_TABLE to first column of TARGET_TABLE

• And there is one direct relation:

  • FROM D of SOURCE_TABLE to first column of TARGET_TABLE

• See Glossary for more info about semidirect relation

Indirect lineage example

Consider this SQL command:

INSERT INTO TARGET_TABLE SELECT COLUMN1 FROM SOURCE_TABLE WHERE COLUMN2 = 'VALUE'

• There is an indirect relation FROM COLUMN2 of SOURCE_TABLE to TARGET_TABLE

Note

• The direct relation connects column endpoint to column endpoint, whereas the indirect relation connects column endpoint to table endpoint
• Direct relation means, that data is copied or transformed from source column and saved into target column.
• Indirect relation means, that target doesn't neccessarily contain the source data, but its presence is somehow influenced by the source column - typically by filtering and aggregating by its value.