Received: from malur.postgresql.org ([217.196.149.56]) by arkaria.postgresql.org with esmtps (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.96) (envelope-from ) id 1wc0o3-0031Lx-2w for pgsql-admin@arkaria.postgresql.org; Tue, 23 Jun 2026 13:04:48 +0000 Received: from localhost ([127.0.0.1] helo=malur.postgresql.org) by malur.postgresql.org with esmtp (Exim 4.96) (envelope-from ) id 1wc0o1-00BsAb-1s for pgsql-admin@arkaria.postgresql.org; Tue, 23 Jun 2026 13:04:45 +0000 Received: from magus.postgresql.org ([2a02:c0:301:0:ffff::29]) by malur.postgresql.org with esmtps (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.96) (envelope-from ) id 1wc0o0-00BsAQ-1m for pgsql-admin@lists.postgresql.org; Tue, 23 Jun 2026 13:04:45 +0000 Received: from resqmta-h2p-567063.sys.comcast.net ([2001:558:fd02:2446::b]) by magus.postgresql.org with esmtps (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.98.2) (envelope-from ) id 1wc0nu-00000001vIa-1ATZ for pgsql-admin@lists.postgresql.org; Tue, 23 Jun 2026 13:04:41 +0000 Received: from resomta-h2p-554994.sys.comcast.net ([96.102.179.205]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 256/256 bits) (Client did not present a certificate) by resqmta-h2p-567063.sys.comcast.net with ESMTPS id bymHwpICajfJ8c0ngwlN3v; Tue, 23 Jun 2026 13:04:24 +0000 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=comcast.net; s=20190202a; t=1782219865; bh=qPpcm++8KeoFAjxR8MGo3wPXGqLdTt0CbzRh+z4a99Y=; h=Received:Received:Date:From:To:Message-ID:Subject:MIME-Version: Content-Type:Xfinity-Spam-Result; b=PoMqmGpYzPglGgvfx2CmxuKBx7AgDUBuhH5ox+ezL1ZZ6fD8PoU/nQAKDF7py40Jz iIihKjWOWLjblLOu5SNKORi6bLomHkW1spc/R1GuHyZ1UwY9rTKKBf/pJwHPJ69DDE c4/x32BVZfUi4Qgnq1ytikiVhv3aBCh3jDlzedjSM8/Zy8odP7jqBy+QXuUBZW1PsV 1xiGdZDlmGtWpC6VDnv+v/r1Yfbq8URo76IDXinwN1EuxnXVar4DrMQwGvh2/DwR8B 2Obze4vEhkepP9nSw1GrCdRS4rME7BIMw50mWvNpCJ0kblEDd3vGvvp0jOx08du2/q 9uT4CKMVHsbcg== Received: from oxapp-ch2f-48o.email.comcast.net ([96.117.85.66]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 256/256 bits) (Client did not present a certificate) by resomta-h2p-554994.sys.comcast.net with ESMTPS id c0nfwIGtMNRCTc0ngwReSm; Tue, 23 Jun 2026 13:04:24 +0000 Xfinity-QID: c0nfwIGtMNRCTc0ngwReSs Xfinity-Graffiti: 49008790502 Xfinity-Graffiti: 52107000864 Date: Tue, 23 Jun 2026 08:04:23 -0500 (CDT) From: JOHN WIENCEK To: bertrand HARTWIG , Ron Johnson Cc: Pgsql-admin Message-ID: <1116567729.67464.1782219863407@connect.xfinity.com> In-Reply-To: References: <359C3FF2-C038-4237-A361-4DAEFA92C114@gmail.com> Subject: Re: About asynchronous I/O MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_67463_547549293.1782219863396" X-Priority: 3 Importance: Normal X-Mailer: Open-Xchange Mailer v7.10.6-Rev83 X-Originating-IP: 2600:1702:7740:6280:a5a6:1a1c:13c0:c884 X-Originating-Port: 32390 X-Originating-Client: open-xchange-appsuite X-CMAE-Envelope: MS4xfLeZX+koflEAzUNjYByMv9e6Fslk5MNLpG0/2bryJ2YC7OSgWj0IyBJv49xNB2vPP5ktQLHJDe0fPfbLjVwMyqQC/BvfXuhnPOjTQRFq80pS5aQefYfb WA8cJpGJSVJnUt3tKBdmAh0b/2yv+4sRxi+bILZ9QVhs6R4pPPj94+9xzbgbZ9L0lhashqpnNLiX6t68SA4eulBDJTb+DIaRPZXdl6GyycdpHfZ1MWvHZ4iV o+t0zGyRSdUv5sbQm/ldrJnIs47tJpxlCZKX1Yoa9HirHycH8+rFohuhZV4MvNKCv0Nz3DrOm4MVPxZy4qy2Xg== List-Id: List-Help: List-Subscribe: List-Post: List-Owner: List-Archive: Archived-At: Precedence: bulk ------=_Part_67463_547549293.1782219863396 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Can your application assume the risk of lost transactions in the event of a= database crash? ASYNC/IO is not a good solution for all applications. Regards John > On 06/23/2026 7:19 AM CDT bertrand HARTWIG w= rote: > =20 > =20 > Very usefull on SSD AND nvme ! (low latency on IOPs). But the higher the = disk latency, the greater the gains you will notice. > =20 > On SSD i observed 200% >=20 >=20 > > Le 23 juin 2026 =C3=A0 14:02, Ron Johnson a = =C3=A9crit : > > On Tue, Jun 23, 2026 at 7:26=E2=80=AFAM bertrand HARTWIG wrote: > >=20 > > >=20 > > > Hello, > > >=20 > > > The size of the database is clearly not important. > > >=20 > > > What matters are queries that are not already cached and that perform= I/O. > > >=20 > > > With async I/O, you can see I/O performance improvements of up to 200= %=E2=80=93300%. The higher the disk latency, the greater the gains you will= notice. > > >=20 > > =20 > > So, not so useful on SSD/NVMe? > > =20 > >=20 > > >=20 > > > Regards, > > >=20 > > > Bertrand > > >=20 > > >=20 > > > > Le 23 juin 2026 =C3=A0 09:55, ek ek a =C3=A9crit : > > > >=20 > > > > Hello everyone, > > > > Are any of you running PostgreSQL 18 on production environments siz= ed between 1 to 3TB? Does the 'asynchronous I/O (AIO) subsystem' deliver a = significant performance increase? Also, has anyone had the opportunity to b= enchmark it against v17? > > > >=20 > > >=20 > > =20 > > =20 > > -- > > Death to , and butter sauce. > > Don't boil me, I'm still alive. > > lobster! > >=20 >=20 ------=_Part_67463_547549293.1782219863396 MIME-Version: 1.0 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable =20 =20
Can your applic= ation assume the risk of lost transactions in the event of a database crash= ?
=20
ASYNC/IO is not a good solution for all applications.


=20
Regards

=20
John
=20
=20
=20
On 06/23/2026 7:19 AM CDT bertrand HARTWIG <hartwig.bertrand@gmail.c= om> wrote:
=20
 
=20
 
Very usefull on SSD AND nvme ! (low latency on IOPs). But the hig= her the disk latency, the greater the gains you will notice.=20
 
=20
On SSD i observed 200%
=20

=20
=20
Le 23 juin 2026 =C3=A0 14:02, Ron Johnson <ronljohnsonjr@gmail.co= m> a =C3=A9crit :
=20
=20
=20
On Tue, Jun 23, 202= 6 at 7:26=E2=80=AFAM bertrand HARTWIG <hartwig.bertrand@gmail.com> wrote:
=20
=20
=20
=20

Hello,

=20

The size of the database is clearly not important.

=20

What matters are queries that are not already cached and that= perform I/O.

=20

With async I/O, you can see I/O performance improvements of u= p to 200%=E2=80=93300%. The higher the disk latency, the greater the gains = you will notice.

=20
=20
=20
 
=20
So, not so useful on SSD/NVMe?
=20
 
=20
=20
=20

Regards,

=20

Bertrand

=20
=20

=20
=20
Le 23 juin 2026 =C3=A0 09:55, ek ek <livadidrive@gmail.c= om> a =C3=A9crit :
=20
=20
=20
=20
Hello everyone,
=20
Are any of you running PostgreSQL 18 on production environ= ments sized between 1 to 3TB? Does the 'asynchronous I/O (AIO) subsystem' d= eliver a significant performance increase? Also, has anyone had the opportu= nity to benchmark it against v17?
=20
=20
=20
=20
=20
=20
=20
=20
=20
 
=20
 
--
=20
=20
Death to <Redacted>, and butter sauce.=20
Don't boil me, I'm still alive.
=20
=20
<Redacted> lobster!
=20
=20
=20
=20
=20
=20
=20
=20
=20
=20
------=_Part_67463_547549293.1782219863396--