Argus

Description

A simple rate limiter service written in Golang. This was built as a quick exercise to revise concurrency in Go. The rate limiter uses a thread-safe BST to store records. A 'shadow' AVL tree is periodically swapped with the BST to provide eventual log(n) guarantees for tree accesses.

Known Issues

• [RESOLVED] Occassionally under heavy concurrent requests, a Read Lock is not being correctly released leading to starvation of the switchover go routine that handle tree swapping.

Todos

• [DONE] Add a function to periodically remove stale records
• The metric used for triggering a tree swap is pretty dubious, so that needs a rethink
• Perform profiling and implement optimizations (e.g. custom json decoding among others)
• • go tool pprof -http=:8000 'http://localhost:6060/debug/pprof/profile?seconds=60'
• Implement an alternate DB engine using perhaps a thread-safe hash table.
• And lastly, there's a bit of cleanup/refactoring needed

Performance Benchmarks

I used Jmeter to exercise Argus on an 8-core, 16G Macbook. There's no connection sharing/reuse among threads in Jmeter. The request body:

{
    "key": "${__RandomString(20,abcdefghijklmnopqrstuvwxyz,)}",
    "capacity": 1000,
    "interval": 60,
    "unit": "s"
}

uses a random string for each request. This obviously affects the lifetime of each record, leading to early evictions during the running of these tests.

Test 1

10 Threads, 50 000 Iterations GOMAXPROCS=8 No Connection Pooling

Test 2

100 Threads, 50 000 Iterations GOMAXPROCS=8 No Connection Pooling

After introducing a function to remove stale records memory usage is hovering at around 3.2 Mb. Obviously this is influenced by the early record evictions in this suite of tests.

Example Usage

curl -X POST -H "Content-Type: application/json" -d '{
    "key": "my_key",
    "capacity": 1,
    "interval": 5,
    "unit": "s"
}' http://localhost:8123/api/v1/limit

Installation

1. Clone the repository.
2. Set environment variables
3. make all
4. ./bin/argus

Alternately use the provided Dockerfile to deploy a containerized version.

$ docker-compose up --build

Contributing

Contributions are welcome. If you would like to contribute please follow these steps:

1. Fork the repository
2. Create a new branch
3. Make your changes
4. Submit a pull request

License

This project is licensed under the Apache License.