Inquiry Regarding Parallel DML (Write-Side) Support in Future PostgreSQL Releases

From: Samil C <samillcinar(at)gmail(dot)com>
To: pgsql-hackers(at)postgresql(dot)org
Subject: Inquiry Regarding Parallel DML (Write-Side) Support in Future PostgreSQL Releases
Date: 2026-05-22 06:53:49
Message-ID: CABp-5JGTRDZXyXuQiMkYv=PAtTHJePx=uZOF4kC9=mTs4kwypQ@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hello PostgreSQL Developers,

I have been analyzing PostgreSQL's parallel execution capabilities and
noticed that parallel query execution is currently available for
operations such as:

Parallel SELECT
CREATE TABLE AS SELECT (CTAS)
Parallel index creation

However, during standard DML operations (INSERT ... SELECT, UPDATE,
DELETE), execution remains effectively single-threaded, preventing the
underlying scans from taking advantage of parallel workers.

I would like to ask whether there are any ongoing discussions or
development plans regarding true parallel DML support in future
PostgreSQL releases. In particular, I am interested in the long-term
direction for parallel execution of INSERT/UPDATE/DELETE operations
and related write-path optimizations.

While I understand the architectural complexities involved in parallel
writes, the current single-threaded execution model remains a
significant bottleneck for large-scale migrations and heavy ETL
workloads.

Until native parallel DML becomes available, are there any recommended
workarounds or best practices for reducing this bottleneck during
large batch operations?

Thank you for your time and for your contributions to PostgreSQL.

Best regards,
SamilC

Browse pgsql-hackers by date

  From Date Subject
Next Message Chao Li 2026-05-22 07:02:48 Fix pg_get_multixact_stats() members_size calculation
Previous Message solai v 2026-05-22 06:15:25 Re: Available disk space per tablespace