MIME-Version: 1.0
References: <df8a9898-0ff9-4805-abc2-f469670ce92d@exa.co.za>
In-Reply-To: <df8a9898-0ff9-4805-abc2-f469670ce92d@exa.co.za>
From: Greg Sabino Mullane <htamfids@gmail.com>
Date: Mon, 16 Feb 2026 10:34:47 -0500
Message-ID: 
 <CAKAnmmKXs4vpUt-3ML2H4rE4+3MEpHqG-rJVqvLqyaW0rfs3Jg@mail.gmail.com>
Subject: Re: Postgres IO sweet spot
To: Riaan Stander <rstander@exa.co.za>
Cc: pgsql-performance@postgresql.org
Content-Type: multipart/alternative; boundary="0000000000000581b0064af2b4e1"
Archived-At: 
 <https://www.postgresql.org/message-id/CAKAnmmKXs4vpUt-3ML2H4rE4%2B3MEpHqG-rJVqvLqyaW0rfs3Jg%40mail.gmail.com>
Precedence: bulk

--0000000000000581b0064af2b4e1
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Tue, Feb 10, 2026 at 6:13=E2=80=AFPM Riaan Stander <rstander@exa.co.za> =
wrote:

> 3 x Host Servers running Microsoft Storage Spaces in a 3 way mirror
>

That's an expensive way to provide some HA. What's the business
requirement? How does that tie into Postgres? Might be able to do it in
other ways.

but at some point it came to light that the Storage Spaces hardware are all
> mechanical disks with NVME only used for Storage
> Spaces journaling and caching. There are now some discussions of upgradin=
g
> drives to SSD, but my concern is that this is not guaranteed
> to solve the issue. Especially with the 3 way mirror it seems all writes
> will go to the other hosts before returning. So latency is almost
> impossible to remove.
>

Yikes! Yes, SSD would be a big win. It's orders of magnitude faster, and
just removes so many problems.

So now my question. I started running some IO tests using fio,
> pg_test_fsync & pg_test_timing. Before we spend days/months trying to tun=
e
> Postgres settings I'm trying to get some definitive published information
> about what IO numbers I should expect when running plain hardware tests
> with Postgres completely out of the loop.


Sorry, I have no numbers to provide you there, but I cannot imagine any
amount of tuning is going to be as big a win as going to SSD.

Cheers,
Greg

--
Crunchy Data - https://www.crunchydata.com
Enterprise Postgres Software Products & Tech Support

--0000000000000581b0064af2b4e1
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr">On Tue, Feb 10, 2026 at 6:13=E2=80=AFPM R=
iaan Stander &lt;<a href=3D"mailto:rstander@exa.co.za">rstander@exa.co.za</=
a>&gt; wrote:</div><div class=3D"gmail_quote gmail_quote_container"><blockq=
uote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1p=
x solid rgb(204,204,204);padding-left:1ex">3 x Host Servers running Microso=
ft Storage Spaces in a 3 way mirror<br></blockquote><div><br></div><div>Tha=
t&#39;s an expensive way to provide some HA. What&#39;s the business requir=
ement? How does that tie into Postgres? Might be able to do it in other way=
s.</div><div><br></div><blockquote class=3D"gmail_quote" style=3D"margin:0p=
x 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">bu=
t at some point it came to light that the Storage Spaces hardware are all m=
echanical disks with NVME only used for Storage <br>
Spaces journaling and caching. There are now some discussions of upgrading =
drives to SSD, but my concern is that this is not guaranteed <br>
to solve the issue. Especially with the 3 way mirror it seems all writes wi=
ll go to the other hosts before returning. So latency is almost <br>
impossible to remove.<br></blockquote><div><br></div><div>Yikes! Yes, SSD w=
ould be a big win. It&#39;s orders of magnitude faster, and just removes so=
 many problems.</div><div><br></div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding=
-left:1ex">So now my question. I started running some IO tests using fio, p=
g_test_fsync &amp; pg_test_timing. Before we spend days/months trying to tu=
ne Postgres settings I&#39;m trying to get some definitive published inform=
ation about what IO numbers I should expect when running plain hardware tes=
ts with Postgres completely out of the loop.</blockquote><div><br></div><di=
v>Sorry, I have no numbers to provide you there, but I cannot imagine any a=
mount of tuning is going to be as big a win as going to SSD.</div><div><br>=
</div><div>Cheers,</div></div><div dir=3D"ltr" class=3D"gmail_signature"><d=
iv dir=3D"ltr"><div>Greg</div><div><br></div><div>--</div><div>Crunchy Data=
 - <a href=3D"https://www.crunchydata.com" target=3D"_blank">https://www.cr=
unchydata.com</a></div><div>Enterprise Postgres Software Products &amp; Tec=
h Support</div><div><br></div></div></div></div>

--0000000000000581b0064af2b4e1--