[Task]: Problem with transaction pipeline

[bykhovets]

What needs to happen?

Hi,

In the example on the screenshot, the insertion into the database is performed one insert per row.

INSERT INTO "BI2_ACCRUAL_CUBE_6860" ("ITEM_ID", "ETOF_ID", "ISD_ID", "CARRIER_ID", ... "COST_USD")
VALUES ( '111111111111'::Integer, '111111111111'::Integer, '111111111111'::Integer, .... '1.11111111111111111111')
...........
INSERT INTO "BI2_ACCRUAL_CUBE_6860" ("ITEM_ID", "ETOF_ID", "ISD_ID", "CARRIER_ID", ... "COST_USD")
VALUES ( '2222222222222'::Integer, '2222222222222'::Integer, '2222222222222222'::Integer, .... '2.222222222222222')
and so on 35 thousand times in our example.

The transaction pipeline ignores batch insertion. And this is correct, since there should be one commit at the end.
But for the Vertika DB, inserting each row as a separate insert leads to the creation of a huge number of files and an error: "to many ROS containers". What makes it impossible to use transaction pipeline with Vertica DB.
Can you suggest any solution that will allow transaction pipelines to work normally with the Vertica database?

Issue Priority

Priority: 3

Issue Component

Component: Database, Component: Pipelines