Content-Type: text/plain;
	charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3776.700.51.11.1\))
Subject: Re: SQL Server's WITH (NOLOCK) equivalent in PostgreSQL?
From: Christophe Pettus <xof@thebuild.com>
In-Reply-To: <CAH0WOV4JhR3OgjNg69h-ZY9o1ooWq=32aa4GrSL5a3y41nApHQ@mail.gmail.com>
Date: Sun, 30 Mar 2025 22:02:18 -0700
Cc: pgsql-general@lists.postgresql.org
Content-Transfer-Encoding: quoted-printable
Message-Id: <C53D7011-110C-4C78-A293-76355C1B2654@thebuild.com>
References: <CAH0WOV4JhR3OgjNg69h-ZY9o1ooWq=32aa4GrSL5a3y41nApHQ@mail.gmail.com>
To: =?utf-8?B?7J207ZiE7KeE?= <jemie9812@gmail.com>
Archived-At: <https://www.postgresql.org/message-id/C53D7011-110C-4C78-A293-76355C1B2654%40thebuild.com>
Precedence: bulk


> On Mar 30, 2025, at 21:44, =EC=9D=B4=ED=98=84=EC=A7=84 =
<jemie9812@gmail.com> wrote:
> Since PostgreSQL uses MVCC, I'm wondering what the best practice is =
for non-blocking reads,=20
> and whether there's an equivalent to dirty reads or READ UNCOMMITTED.

There are two different questions here.

1. Reads are not blocked in PostgreSQL by writers, even if another =
transaction has updated the row but not yet committed.

2. There is no equivalent of a dirty read in PostgreSQL.  You can only =
read committed data (short of bypassing the database and reading the =
disk directly).=