Ask HN: JMeter Alternative? | Hacker News Reader

Ask HN: JMeter Alternative?

5 pointsfebrilian8 comments

What's the simplest app to measure requests per second like JMeter? I'm looking for go-based or rust-based apps. Thanks

I have not used it myself but colleagues of mine recommend k6: https://k6.io/open-source/

There is also a Github repository keeping track of all the load-testing tools out there: https://github.com/denji/awesome-http-benchmark You will probably find something that will fit your bill.

I do not understand though why it is of any relevance what language the application is written in. Yes, Go and Rust indicate a certain speed and novelty but you can write fast applications with good UX in other programming languages too (and bad ones in Go and Rust).

That k6 app looks sleek, I'll try it out. Thanks!

Yes I don't mean to dismiss all other languages I just happen to only have these two runtimes and few others and I'm currently having problems with Java runtime which result in me not being able to run JMeter. Plus, if it's open source it can benefit me because these are the two languages I practically use.

Unlike Java applications, apps written in Go and Rust do not need a runtime installed on your computer. The same is true for C, C++ and a lot of other languages. You are most probably less constrained than you think.

You're right, thanks for explaining

Yeah, k6 is great, super easy to use for programmers.

+1 for K6. The open source is quite easy to setup and their cloud (freemium) makes life pretty easy to scale/distribute/visualize tests. I use it for https://wphostingbenchmarks.com for many years now (used to be Load Impact).

I've used LoadForge before for stress testing:

https://loadforge.com

I found it a good middle-ground between DIY tools like "hey" and the likes of JMeter and K6.

LoadForge is really just a frontend for Locust [2] behind the scenes so all tests are written in Python which might not fit your requirement for Go/Rust, but it's affordable and quick to get started with.

[1] https://github.com/rakyll/hey [2] https://locust.io

I suppose the end goal is to replicate production traffic patterns as close as possible. Why not just use production traffic? Of course omitting PII is mandatory.

Take a look at goreplay. https://github.com/buger/goreplay/wiki

Reply on news.ycombinator.com