RE: AIX support

Hi Tom and community members,

Firstly, thank you for all the time, guidance, and testing the patches so far.

>> Remind me again why anyone would choose to run Postgres on this
>> platform? Why are we moving mountains to make it possible?

AIX users are actively consuming open-source tools. PostgreSQL is one of them and enabling support is essential for them to adopt modern database workloads on the platform. This includes not only the core PostgreSQL engine but also extensions such as pgvector, which are for AI, search, and analytics use cases.

To make this possible, AIX has been contributing across open‑source communities like Meson, CMake, GDB, ONNXRUNTIME, and others.

Our goal is to ensure that AIX users have access to the modern tools available in the open-source world. This is why we want to work closely and consistently with the PostgreSQL community.

As part of the efforts, we recently contributed a patch in the below link for PostgreSQL to work well with meson without exporting OBJECT_MODE=64

https://github.com/mesonbuild/meson/pull/15516

So, we are very motivated to be part of the PostgreSQL community and have AIX supported in the PostgreSQL code.

>Unless there is something seriously wrong with how cfarm119 is set up,
>the conclusion has to be that AIX is mind-bogglingly bad at disk I/O.

We reviewed the configuration of cfarm119 and found that it is running as a shared LPAR using vSCSI‑backed disks. The filesystem under /home is shared with other communities as well, including the CMake community, so multiple workloads are accessing the same storage concurrently. This shared and virtualized setup can significantly affect I/O performance. Our teams are continuously investigating the setup to understand the issue.

For more representative performance measurements, we recommend using the douc node instead. Let us know and we will provide you the access to this node.

>On 04/02/26, 5:33 AM, "Michael Paquier" <michael(at)paquier(dot)xyz> wrote:
>Saying that, I'm really glad for the report that has led to the
>discovery fixed in d46aa32ea5ce. Opportunistic improvements not
>specific to AIX are also good to know about: they show long-term
>benefits everywhere.

Thanks for that, and I’m really glad the investigation led to fixes that benefit all platforms.

We greatly appreciate all the feedback and guidance so far, and we look forward to continuing as active contributors to the PostgreSQL and AIX communities.

Have a nice day,

Best regards,

Aditya.