A light weight loadbalancer for canary deployments. This utilizes the Weighted Round Robin algorithm to balance requests across resources based on defined weight.

Weights are defined with whole numbers which represents nx10, and the total weight of all servers added to the loadbalancer must sum up to 10.

Example

weight 5 represents 50% of traffic sent to the server
weight 3 represents 30% of traffic sent to the server
weight 2 represents 20% of traffic sent to the server

Total weights sums up to 10 which represent 100% of all traffic

Getting Started

You can either run the loadbalancer with docker or clone the repo, build and run the binary with the docker method being the easiet.

An important requirement is to have a config.json to load the servers and other details to the loadbalancer. Below is an example of the config file.

{
  "healthCheckInterval": "2s",
  "servers": [
    {
      "url": "http://localhost:5100",
      "weight": 5,
      "healthCheckUrl": "http://localhost:5100/health"
    },
    {
      "url": "http://localhost:5200",
      "weight": 3,
      "healthCheckUrl": "http://localhost:5200/health"
    },
    {
      "url": "http://localhost:5300",
      "weight": 2,
      "healthCheckUrl": "http://localhost:5300/health"
    }
  ]
}

Docker setup

Requirements

• Docker installed
• config.json file that will be mounted to the docker container

Run the command

docker run -p 7080:7080 -v ./config.json:/app/loadbalancer/config.json --name loadbalancer -d ghcr.io/vickysomtee/loadbalancer

Clone repository

Requirements

• Go installed
• config.json file

Clone the Repository to your local directory

git clone https://github.com/Vickysomtee/loadbalancer.git

Build an executable

cd loadbalancer
go build

Run the executable

./loadbalancer -config=config.json

Note: If you created your config.json file in the current working directory, there is no need to specify the config argument. Run the executable using

./loadbalancer

Contributing

Contributions are welcome! Please submit a pull request or create an issue to contribute to this project.