Redis Proxy

Caching Redis proxy with HTTP and Redis API.

Tryout HTTP API:

docker-compose up -d redis app
docker-compose exec redis redis-cli SET hello world
curl $(docker-compose port app 80)/values/hello

Tryout Redis API:

docker-compose up -d redis app
docker-compose exec redis redis-cli SET hello world
redis-cli -p $(docker-compose port app 6379 | cut -d : -f 2) GET hello

The solution is written in Python 3.6.

Features

Restful HTTP API
Limited Redis API (only GET and PING)
Asynchronous I/O
Redis backend connection pooling
Supports both string and binary keys
Health check

Configuration

The proxy is configured via environment variables:

Name	Description
LISTEN_PORT	Port the HTTP API listens on (default: 80).
WORKERS	Number of worker processes (default: 1).
REDIS_CONNECT	Redis connection string (e.g.: `redis://localhost:6379`).
REDIS_DB	Redis database index to connect (default: 0).
MAX_CONCURRENCY	Maximum concurrent connections to the Redis backend (default: 1).
CACHE_CAPACITY	Cache capacity.
CACHE_TTL	Cache global ttl.
REDIS_SERVER_LISTEN_PORT	Port the Redis API listens on (default: 6379). Set 0 to disable.
REDIS_SERVER_TIMEOUT	Redis API command timeout (default: 1s).

HTTP API

The proxy exposes HTTP Restful API for getting values.

It supports both string keys:

GET /values/<string_key>;

and binary keys as base64 encoded strings (b=1 query parameter must be set):

GET /values/<base64_encoded_binary_key>?b=1

Cache

LRU cache with global expiration.

Read and write complexity is O(1).

Tests

Prerequisites:

docker 17.09.0+
docker-compose 3.4+

Use this command to run tests:

make test

The command runs tests using docker-compose.

It takes about a minute to build everything and run all the tests.

The test suite has both integration and unit tests.

Integration tests are relying on docker-compose to spin up Redis instance.

TODO

Prometheus metrics endpoint
Structured logging in JSON format
Cache control headers

SHSE / redis-proxy