This repo has a lot of services so a natural progression is to reconsider the implementations such that people can clone the repo and run some benchmark.
What is that benchmark and how should it be run? No idea atm. Some may argue there are more efficient ways to reduce allocations, use more efficient libraries for (de)serialization, maybe even requests through docker may be frowned upon but if it's all in docker then it's relative.
We'd probably need to run 1 service at a time and maybe even restart the generator between each run.
A half-assed benchmark is pointless, and micro benchmarking isn't ideal but this project should allow ppl to run something or easily extend it to run proper benchmarks.
Note: remember the project has telemetry which will affect performance so we should look into running it with(out) too.
This repo has a lot of services so a natural progression is to reconsider the implementations such that people can clone the repo and run some benchmark.
What is that benchmark and how should it be run? No idea atm. Some may argue there are more efficient ways to reduce allocations, use more efficient libraries for (de)serialization, maybe even requests through docker may be frowned upon but if it's all in docker then it's relative.
We'd probably need to run 1 service at a time and maybe even restart the generator between each run.
A half-assed benchmark is pointless, and micro benchmarking isn't ideal but this project should allow ppl to run something or easily extend it to run proper benchmarks.
Note: remember the project has telemetry which will affect performance so we should look into running it with(out) too.