This simple project was meant to be an opportunity to play with a handful of enterprise technologies. It is a reverse proxy written in Go with support for Redis caching, Prometheus metrics and distributed tracing.
After cloning the repo run docker-compose build && docker-compose up to build and start the application with all its supporting services. Exercise the reverse proxy by querying the application through it:
$ curl http://localhost:8080/api/fibonacci?pos=10000
2132534333
$ curl http://localhost:8080/api/echo?echo=test
test
Navigate to the Jaeger UI at http://localhost:16686/ to see some traces:
Prometheus metrics are available in the usual spot:
$ curl http://localhost:8080/metrics
# HELP go_rev_proxy_requests_latency_ms Summary of Request Latency (ms)
# TYPE go_rev_proxy_requests_latency_ms summary
go_rev_proxy_requests_latency_ms{path="/api/echo",quantile="0.5"} 1
go_rev_proxy_requests_latency_ms{path="/api/echo",quantile="0.9"} 11
go_rev_proxy_requests_latency_ms{path="/api/echo",quantile="0.99"} 11
go_rev_proxy_requests_latency_ms_sum{path="/api/echo"} 15
go_rev_proxy_requests_latency_ms_count{path="/api/echo"} 6
go_rev_proxy_requests_latency_ms{path="/api/fibonacci",quantile="0.5"} 1
go_rev_proxy_requests_latency_ms{path="/api/fibonacci",quantile="0.9"} 300
go_rev_proxy_requests_latency_ms{path="/api/fibonacci",quantile="0.99"} 300
go_rev_proxy_requests_latency_ms_sum{path="/api/fibonacci"} 303
go_rev_proxy_requests_latency_ms_count{path="/api/fibonacci"} 6
Logging shows the impact of caching
proxy_1 | Starting Request http://backend:8080/api/echo?echo=test
proxy_1 | Caching: Miss http://backend:8080/api/echo?echo=test
proxy_1 | Ending Request http://backend:8080/api/echo?echo=test
proxy_1 | 2019/07/10 02:39:58 HTTPRequest took 11.4113ms
proxy_1 | Starting Request http://backend:8080/api/echo?echo=test
proxy_1 | Caching: Hit http://backend:8080/api/echo?echo=test
proxy_1 | Ending Request http://backend:8080/api/echo?echo=test
proxy_1 | 2019/07/10 02:39:58 HTTPRequest took 1.0331ms
Overall this turned into a fun project that allowed for some basic exploration of go, caching, tracing and other technologies. This code should never be used in any real world scenario. However, if you're interested in doing similar things you may find some useful sample code buried in here.
Improvements that could be made:
- Respect incoming OpenTracing headers
- Set outgoing OpenTracing headers
- Add distributed rate limiting through Redis
- Make caching, tracing, etc. configurable
- Only cache gets
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent