[ATLAS-1207] Dataset exists query in lineage APIs takes longer

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.8-incubating
• Assignee: Shwetha GS
• Reporter: Sharmadha S
• Created: 04/Oct/16 10:34
• Updated: 21/Oct/16 06:07
• Resolved: 21/Oct/16 06:07

Description

Hive_column now extends DataSet. Lineage Service uses the DSL query Dataset where _guid = <id> which maps to the gremlin query g.V().has(supertype, Dataset).has(_guid, <id>). Since the first filter is on type which returns many vertices, this query is slow. Supertypes is a list property and not sure how adding combined index will work. This can be replaced with graph query directly like

titanGraph.query().has(Constants.GUID_PROPERTY_KEY, guid)
                          .has(Constants.SUPER_TYPES_PROPERTY_KEY, AtlasClient.DATA_SET_SUPER_TYPE)

Thanks ssainath for helping to test this