Message-ID: <bd1aeac0-942e-4e2f-9e9e-034eac610a70@proxel.se>
Date: Mon, 6 Apr 2026 16:59:42 +0200
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: Linux 7.0 performance degradation
To: Rick Otten <rottenwindfish@gmail.com>,
 "pgsql-performa." <pgsql-performance@postgresql.org>
References: 
 <CAMAYy4+ffiH0dczV3Wyv9jx3hNSR_Wf2t0Y+1j1YJizEbQzVJA@mail.gmail.com>
From: Andreas Karlsson <andreas@proxel.se>
Content-Language: en-US
In-Reply-To: 
 <CAMAYy4+ffiH0dczV3Wyv9jx3hNSR_Wf2t0Y+1j1YJizEbQzVJA@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/bd1aeac0-942e-4e2f-9e9e-034eac610a70%40proxel.se>
Precedence: bulk

On 4/6/26 4:12 PM, Rick Otten wrote:
> I read this article of the weekend which was reporting that the latest 
> Linux kernel kills postgresql performance:
> https://www.phoronix.com/news/Linux-7.0-AWS-PostgreSQL-Drop <https:// 
> www.phoronix.com/news/Linux-7.0-AWS-PostgreSQL-Drop>
> 
> I haven't seen any discussion here about this yet and was wondering what 
> your thoughts are?  Are we doing to have to upgrade our postgresql 
> versions before we can upgrade our Linux kernel?  Are there any hacks or 
> tricks or workarounds to help mitigate this issue?
> 
> Is it real?

I recommend reading the linked thread[1] at LKML where Andres determined 
that the regression is likely only present when you do not use huge 
pages. So as long as you make sure that you use huge pages, which you 
should anyway, there is no need to worry about this regression. You can 
also read some discussion at Hacker News where Andres () is also 
involved in the discussion[2].

1. 
https://lore.kernel.org/lkml/xxbnmxqhx4ntc4ztztllbhnral2adogseot2bzu4g5eutxtgza@dzchaqremz32/
2. https://news.ycombinator.com/item?id=47644864

Andreas