Show HN: I made a platform to load and performance test API's | Hacker News Reader

Show HN: I made a platform to load and performance test API's

buzzbench.io

30 pointslazarkap14 comments

I built BuzzBench because I was frustrated with how complex performance testing tools have become. And I was ending up writing my own scripts to test endpoints and manually check out resource usage at the time of testing.

Checkout demo: https://www.youtube.com/watch?v=yAnbZMoQvmQ

Open articleView on HN

This is a pretty crowded space, with a wide range of solutions for all kinds of situations. Have you looked into https://k6.io/ ? IMO it strikes a good balance of being simple to start with, but flexible to support advanced use cases if needed.

At the very least, the first thing I would like to see is a comparison of your platform to existing tools and services.

Thanks for this comment, here is something fast I pulled around:

Distributed Testing: BuzzBench runs agents directly in your environments, while Gatling/k6 typically run from a single location

Coding Required: BuzzBench uses UI configuration, while Gatling needs Scala and k6 needs JavaScript

Dashboard: BuzzBench includes a built-in dashboard, Gatling/k6 require separate setups

Environments: BuzzBench tests multiple environments simultaneously, harder to coordinate with Gatling/k6

Setup: BuzzBench has a simpler install with lightweight agents vs. more complex configuration for others

Learning Curve: BuzzBench is designed for quick adoption, Gatling/k6 have steeper learning curves

I’ve been looking for something just like this. And thinking of writing my own just like you ;) I’m new to the whole web app area, coming from desktop: how would you best explain or pitch this to a newbie who understands the need but that’s about it?

Can it be used for once-off testing (per deployment we can handle n thousand connections / hour) or can it track performance and response speed so you can see how your servers are doing over time?

So BuzzBench can basically be a health check for your web app, it hammers your endpoints with requests to see if they hold up under pressure.

You can definitely use it for both quick "will this break under load" tests before deploying, and to track performance over time so you can catch when things start getting sluggish.

The cool part is you can run tests from pretty much anywhere, your laptop, your CI pipeline, different regions - so you get realistic results rather than some theoretical benchmark.

Few years ago, I had to load test production site for upcoming events. Local test was unrealistic. Some testing tool was re-using connection. I used loader.io to test. There was some issues I can't recall but it was great for both frontend and backend. I was able to test frontend at around 10 gbps.

Looks neat, but I hate the "limited free" thing. What's the actual price? If I make this product a part of my team's workflow, what's the future sticker shock?

The middle tier says "€5".

I agree, my initial plan was to get users to actually use it for free, and when I think I offer enough value, put a price on it, but of course, I would not be like "from tmrw, you pay this amount", it would be months in advance.

Interesting. Internally we made a similar tool for performance testing applications, including fault and latency simulation of dependencies. It also includes analysis of resource consumption and whether performance and/or resource consumption have increased or decreased.

This is now the second time a similar tool has popped up here at HN, maybe we should open source it.

There is an open source Im working on, that will be able to achieve all this just without dashboard https://github.com/LazarKap/buzzbench.io. How did you analyze resource consumption with your tool??

Yes that was the “second” one I was referring to.

We just piggyback on the Prometheus metrics provided by cadvisor for the container level metrics. Then for the jvm we collect metrics such as memory allocation etc.

This all goes into a big table and the user can select the right version. Next iteration will be where they can configure whether to aim for latency, throughput, or efficiency, and which weights so we can select the winner automatically.

I plan to add ability to deploy additional agent to the env that is being tested, and that one would just grab resource metrics and send back to me while its being tested. :D So you can see your resource usage while tests are running and you can predict how much traffic/usage your platfroms/services can handle

How is it any different than something like Gatling?

Hey, imo, its much easier to setup, and its running tests FROM different environments, I think gatling is standalone tool and it is running test from their environment?

Reply on news.ycombinator.com