benchmark.js VS Gatling

We have a series of benchmark suites using https://benchmarkjs.com/. Every CI run saves the results of the benchmarks to a json file that we persist across runs and can compare against.

We use https://benchmarkjs.com/ to get statistically significant results, then write the results with the git hash to get a view of performance after every CI build.

We used Benchmark.js for benchmarking and piscina as a pool for worker threads. benchmark.js was used to run the same code in 2 scenarios - one using a single thread and one using the piscina pool. The degree of parallelism was passed to the program via an environment variable. The test code is present in worker.js in both the cases.

JavaScript: For JavaScript, there are already some good tools for benchmarking, most notable being Benchmark.js and Bench-Rest. Using these tools will allow you to be able to properly test the performance of your code. It is generally given that you want to use software already tested for acceptable benchmarking as the demos shown today are often trivial and may not give all the results you want.

Benchmark.js

https://benchmarkjs.com is what you want.

Gatling: An open-source load and performance testing tool primarily designed for web applications, Gatling utilizes a simple domain-specific language (DSL) for creating and maintaining test scripts. It supports HTTP/2 and allows recording and generation of scenarios directly from a browser. The tool also provides detailed performance reports that are easy to analyze.

Performance and load testing are essential parts of GraphQL API testing. It ensures APIs can handle expected traffic volumes and respond within acceptable timeframes. You can use tools like Apache JMeter or Gatling to generate realistic loads and evaluate the API's performance under different scenarios. Techniques like batched queries and caching can help mitigate this issue.

New to the .NET community and trying to learn! I have used tools such as Apache JMeter (Java), gatling.io (Java) and Locust (Python) that are decent full featured web perf frameworks. Typically these integrate well with your code, and can be run as part of your unit/integration tests and produce offline reports

Gatling , this is what we tested concurrency with. Setting up might take a while depending on your exp. but the tool is solid.

I used SpringBoot 3.0.2, GraalVM 22 (JVM mode), a MacOS 2,6 GHz 6-Core Intel Core i7, running 1000 users for 5 minutes. The idea was to test how memory consumption and CPU usage evolve. Below, I compared the footprint of these three solutions. I collected the total count of requests, throughput, memory consumption, and CPU usage using VisualVM and Gatling.

Load test your code to simulate a production load (you can use tools like k6 or Gatling)

k6 is our preferred javascript based cloud tester for light webservices. gattling if you have (a monolith with) workflows and need more formalized code to prove/audit your testing.

It's read as "Ochikisu" and origin is the machine gun manufacturer Hotchkiss because this tool does not have as much firepower as the Gatling.