Deployment

This page shows how to deploy and test all the backend software developed in this project

Requirements

• Windows OS with powershell
• git bash
• wsl
• Rancher Desktop
• python3

Regarding the deployment of the backends developed with different programming languages ​​and technologies, the execution of the benchmarks and the processing of results, several scripts were created to automate the entire process.

The deployment is made up of 3 parts, the backends launched locally as part of the operating system, the backends in docker as containers and the JTQ migrated to wasm using rust.

Each of these parts can be executed through PowerShell by executing the following scripts respectively (Git Bash needs to be set as the default program to execute files with extension .sh):

• launch_local.ps1. Requirements: node, npm, java (jdk 11), maven, pip, rust, cargo
• launch_docker.ps1
• launch_wasm.ps1. Requirements: subo (wsl)

The process used in the scripts consists in measuring the idle consumption for 5 minutes, then launching all the backends sequentially and, for each one, testing both created benchmarks. Finally a process is applied to the generated HTML files in order to remove unnecessary information (cleaner.py).

Also, to get power consumption for the backend with an edge computing architecture, a service available here must be executed in both servers (central and edge) using cargo. Then, the measurements will be triggered by running and setting the ip of each server:

• launch_cloud_edge_edge.ps1.

Warning: All the scripts were developed and tested on Windows together with Rancher Desktop to use Docker and a WSL environment. They are probably not ready to be run directly on other operating systems.