Re: Asking for feedback on Pgperffarm

From: Mark Wong <markwkm(at)gmail(dot)com>
To: Yedil Serzhan <edilserjan(at)gmail(dot)com>
Cc: pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: Re: Asking for feedback on Pgperffarm
Date: 2022-08-08 17:06:14
Message-ID: YvFChl4y5vFVAvs/@moubin
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi Yedil,

On Mon, Aug 08, 2022 at 02:50:17PM +0200, Yedil Serzhan wrote:
> Dear hackers,
>
> I'm Yedil. I'm working on the project "Postgres Performance Farm" during
> Gsoc. Pgperffarm is a project like Postgres build farm but focuses on the
> performance of the database. Now it has 2 types of benchmarks, pgbench and
> tpc-h. The website is online here <http://140.211.168.145/>, and the repo
> is here <https://github.com/PGPerfFarm/pgperffarm_server>.
>
> I would like you to take a look at our website and, if possible, give some
> feedback on, for example, what other data should be collected or what other
> metrics could be used to compare performance.

Nice work!

We need to be careful with how results based on the TPC-H specification
are presented. It needs to be changed, but maybe not dramatically.
Something like "Fair use derivation of TPC-H". It needs to be clear
that it's not an official TPC-H result.

I think I've hinted at it in the #perffarm slack channel, that I think
it would be better if you leveraged one of the already existing TPC-H
derived kits. While I'm partial to dbt-3, because I'm trying to
maintain it and because it sounded like you were starting to do
something similar to that, I think you can save a good amount of effort
from reimplementing another kit from scratch.

Regards,
Mark

--
Mark Wong
EDB https://enterprisedb.com

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Julien Rouhaud 2022-08-08 17:11:40 Re: Get the statistics based on the application name and IP address
Previous Message Peter Geoghegan 2022-08-08 16:48:18 Re: Remaining case where reltuples can become distorted across multiple VACUUM operations