One of the first things that we think for optimizing api's is caching, but we need be aware of what cache strategy suites each case better.
This repository consist of a load test using Artillery, a open-source load test toolkit, in a simple microservice using in-process and distributed cache strategies aiming to compare and measure latency impact between both.
You should not decide your cache strategy only by measuring latency. Each strategy has it's own drawbacks, for example, despite latency a in-process cache strategy do not suits well for an environment that you need shared cache data. For a deep understading of this subject I recommend the following article.
Some days ago talking with @desk467, he pointed out how a in-process cache strategy could have better performance compared to a distributed cache strategy in some cases. Based on this conversation, I have decided to measure the latency impact on reading operations using node-cache and redis for in-process and distributed strategy respectively in a simplified environment.
First you need npm, docker and docker-compose installed
Next, install Artillery
npm install -g artilleryNext, launch the services
docker-compose up -dNext, make the script for run the load test in docker executable
chmod +x ./scripts/search-messages-docker.shNext, run the load test for each service
./scripts/search-message-docker.sh <service-name>i.e
./scripts/search-message-docker.sh in-process-server
./scripts/search-message-docker.sh redis-server 
The MIT License (MIT)
Copyright (c) 2021 Lucas Machado