taxi HTTP API

Installation

$ go get github.com/Alma-media/taxi

Docker

Build:

$ docker build . -t taxi

Run:

$ docker run --rm -it -p 127.0.0.1:8080:8080 taxi

Further improvements

API

enable the access control for public and private endpoints (use middleware)
enable codec middleware and retrieve codec from the context to use corresponding encoder (e.g. codec.NewEncoder(w).Encode(order))
pass logger to the handler (intermal errors should be logged but not exposed to the user)
find approximate limit to enable throttling

Config

find a better library to parse the values from flags, toml, yaml, json ...

Repository

switch to event driven approach and get rid of proxy (current repository implementation) to increase the throughput

Storage

choose another storage implementation that uses the context and is able to fail / return an error (e.g. database, blockchain etc)

Benchmark results

[alma@alexsch taxi]$ ab -n 1000000 -c 1000 http://localhost:8080/request/

Server Softwa[![GoDoc][godoc-badge]][godoc-link]
re:        
Server Hostname:        localhost
Server Port:            8080

Document Path:          /request/
Document Length:        3 bytes

Concurrency Level:      1000
Time taken for tests:   37.403 seconds
Complete requests:      1000000
Failed requests:        0
Total transferred:      118975648 bytes
HTML transferred:       3000000 bytes
Requests per second:    26735.49 [#/sec] (mean)
Time per request:       37.403 [ms] (mean)
Time per request:       0.037 [ms] (mean, across all concurrent requests)
Transfer rate:          3106.32 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0   18   5.4     17      45
Processing:     4   20   5.8     19      78
Waiting:        0   11   5.6      9      55
Total:         19   37   5.9     35     104

Percentage of the requests served within a certain time (ms)
  50%     35
  66%     36
  75%     38
  80%     39
  90%     47
  95%     51
  98%     55
  99%     58
 100%    104 (longest request)