Thanks, Wang, for initiating this discussion. I think establishing a dedicated set of benchmarks for Fluss is a great idea.
I suggest using the repository name `fluss-benchmarks` — the plural form makes sense as it will host multiple benchmarking frameworks and workloads. Additionally, there is already a module named fluss-benchmark in the apache/fluss repository. To avoid confusion between the module and the new benchmarks repo, I suggest renaming the existing module to fluss-jmh. This would better reflect its purpose as a collection of microbenchmarks using JMH. Best, Jark On Wed, 30 Jul 2025 at 15:23, yuxia <[email protected]> wrote: > Thanks Cheng for driving this work. It's really important for Fluss. For > me, the name `fluss-e2e` is not straight forward. Since it's mainly for > benchmark, > I'd like suggest name it with benchmark related, such as > fluss-benchmark(flink name flink-benchmark), or fluss-bench(clickhosue name > clickbench). > > Best regards, > Yuxia > > ----- 原始邮件 ----- > 发件人: "Wang Cheng" <[email protected]> > 收件人: "dev" <[email protected]> > 发送时间: 星期一, 2025年 7 月 28日 下午 5:51:06 > 主题: [DISCUSS] A new repository for fluss-benchmark > > Hi everyone, > > > I'm writing to propose the creation of a new repository named fluss-e2e. > This repository will house a comprehensive suite of benchmarking tools > designed to simplify performance evaluation of Fluss clusters across > diverse scenarios. > > > For the primary-key table benchmark, I've developed a proof-of-concept > (PoC) implementation [1], which leverages Alibaba Cloud Hologres' > end-to-end performance testing tool [2][3]. For the log table benchmark, we > can adapt the Open Messaging Benchmark (OMB) framework [4] as a > foundational starting point. > > > Importantly, fluss-e2e differs significantly from the existing > fluss-benchmark directory [5] in the current Fluss mainline. The legacy > fluss-benchmark focuses on launching a local pseudo-distributed Fluss > cluster and solely evaluates log tables. In contrast, fluss-e2e aims to > provide a robust, multi-scenario performance testing suite that can > validate any Fluss cluster deployment - whether primary-key or log tables, > local or distributed. > > > Lastly, any ideas for a better name for this repo? > > > Looking forward to your feedback :) > > > [1] https://github.com/xx789633/fluss-e2e-performance-tool > [2] > https://www.alibabacloud.com/help/en/hologres/user-guide/best-practices-for-performance-tests-on-data-writes-data-updates-and-point-queries > [3] > https://github.com/aliyun/alibabacloud-hologres-connectors/tree/master/holo-e2e-performance-tool > [4] https://openmessaging.cloud/docs/benchmarks/ > [5] https://github.com/apache/fluss/tree/main/fluss-benchmark > > > > Regards, > Cheng > > > > >
