redis-cache

README

redis-cache - HTTP JSON caching proxy for Redis

Architecture Overview (what the code does)

• The proxy webserver listens on /redis for GET requests with the param key.
• Up to 100 client requests at once will run using parallel concurrent processing while the rest will be queued. The client connection remains open until their request is processed.
• The client request key/value for the GET request is cached using a fixed capacity, global expiry, LRU eviction cache.
• Less than 100 LOC.

Make and Run

• note: flags are optional

make test
./redis-cache --ttl 360 --keycap 128 --port 8000 --address localhost:6379

Docker

docker-compose up

Complexity of Caching - Big-O

O(1)

Example Client

import(
    "github.com/apibillme/restly"
)

func main() {
    req := restly.New()
    res, statusCode, err := restly.GetJSON(req, "http://localhost:8000/redis", `?key=123`)
    if err != nil {
        if statusCode == 200 {
            value := res.Get("value").String()
            log.Println(value)
        } else {
            value := res.Get("error").String()
            log.Println(value)
        }
    }
}

Load Testing

redis-cache-benchmark

The first key set expires after one second and the test ran for 5 seconds - run against docker-compose in this repo on local Mac.

Total Requests: 5000
Requests Per Second: 1000
Success Ratio: 100%
Max: 183.256182ms
Mean: 15.357077ms
50th percentile: 730.348µs
95th percentile: 99.38549ms
99th percentile: 149.668733ms

Not Implemented

• Redis RPC - took too long to write the parser and dealing with net package and parallel concurrent processing was something I didn't solve.