Load Balancer ⚖️

Load balancer system for the discipline Distributed Systems - UNICAMP

Usage • Options • Requests • Capacity • Metrics

A load balancer simulator handles arriving requests and distributes them between three independent servers.

Usage

To run the project you have two options:

1. Docker (Recommended)

If you have docker, you can build and run the project with simple commands

docker compose build

docker compose run --rm load-balancer

2. Manual

If you prefer you can install rust mannualy and run through cargo

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

cargo run

Options

It is possible to define how the system behaves

Balancing Mode

• Random: The servers are chosen randomly.
• Round Robin: The servers are chosen uniformly, regardless of their workload.
• Smaller Queue: The server with the smallest request queue (i.e. the smallest workload) is chosen.

Arrival Rate (λ)

You can set the average number of requests arriving per second between 0 and 10.

Requests

Requests are defined by type and size.

• Type

  • CPU Bound: It demands more CPU computing.
  • IO Bound: It demands more input/output waiting time.
  • Mixed: A mixture of previous types.

• Size

  • Small: 100ms
  • Mid: 300ms
  • Large: 1000ms

Capacity

Currently, the maximum number of requests that each queue can store is hard-coded. However, you can easily modify this to test new scenarios.

• Server (each): 10 requests
• Pending list: 20 requests

Metrics

• Total Requests: The total number of requests received.
• Processed: Number of successfully processed requests.
• Avarage Response Time: The average time taken to process a request since its arrival.
• Throughput: Number of requests processed per second.