Hi everyone,

I, Yuepeng(CC'ed), Youngil(CC'ed) created FLIP-602[1][2] to introduce two
new scan partitioning options for the JDBC connector.


The current scan.partition.column implementation splits data by equal
numeric intervals, which breaks down under skewed data and offers no
solution when no suitable numeric column exists.


FLIP-602 proposes two extensions:


1. scan.partition.use-physical-id — Partitions data using the database's
native physical row identifier (Oracle ROWID, PostgreSQL ctid), producing
naturally balanced splits with no user-defined partition column required.


2. scan.partition.boundary-query — Accepts a user-provided SQL query that
returns ordered split points, enabling distribution-aware partitioning for
any column type including strings.


The vote will be open for at least 72 hours.


 +1 Approve

  0 No opinion

 -1 Disapprove (please provide a reason)


[1] cwiki.apache.org/confluence/x/x4Q_Gg

[2] https://lists.apache.org/thread/wxh94hhyoy4b164m9q6cxhf8jypkpwdo


Best regards,

Chanhae Oh

Reply via email to