Shwetha G S created ATLAS-442:
---------------------------------

             Summary: Hive model to handle cascaded deletes
                 Key: ATLAS-442
                 URL: https://issues.apache.org/jira/browse/ATLAS-442
             Project: Atlas
          Issue Type: Bug
            Reporter: Shwetha G S


Ideally, hive model should be like this:
type - attributes
hive_table - ..., partitions(list<hive_partition>), ...
hive_partition - values(list<string>), ...

partitions attribute in hive_table is composite. Add partition is frequent in 
hive, and this would result in create new partition and add this reference to 
table. So, this results in create entity + update array for hive_table entity. 
Since the number of partitions can be large, we want to avoid updates to 
hive_table entity every time. 

Hence hive model is implemented now as:
hive_table - ...
hive_partition - table(hive_table), values(list<string>), ...

In this model, there is no way to indicate that partitions are composite 
attributes of table and hence deleting table will nor result in deleting of 
partitions. But we need the cascading delete of partitions in this case. We 
need to figure out a way to solve this.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to