use "wrk" tool to test it for prologue.

[lihf8515]

This is a wrk test of httpbeast hello word：wrk -c1000 -d10s http://192.12.1.22:8080/

Running 10s test @ http://192.12.1.22:8080/
2 threads and 1000 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 8.01ms 12.16ms 215.10ms 99.05%
Req/Sec 71.46k 5.27k 76.81k 81.31%
1407514 requests in 10.05s, 154.37MB read
Non-2xx or 3xx responses: 1407514
Requests/sec: 140036.79
Transfer/sec: 15.36MB

This is a wrk test of prologue hello word.:wrk -c1000 -d10s http://192.12.1.22:8080/

Running 10s test @ http://192.12.1.22:8080/
2 threads and 1000 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 34.58ms 0.96ms 39.56ms 78.50%
Req/Sec 14.51k 1.18k 16.28k 85.35%
286065 requests in 10.02s, 43.65MB read
Requests/sec: 28538.37
Transfer/sec: 4.35MB

Why is the difference so big？which place cause it so low, suggest to optimize.

[ringabout]

I haven't used valgrind to profile it.
But I guess there may exists some aspects which caused its slowness:

1. httpbeast uses nil HttpHeaders default, Prologue uses empty HttpHeaders default.
2. Prologue uses middlewares which increases overhead.
3. Prologue parses query parameters, form parameters and so on.
4. It has startup, shutdown event, static file, to check whether exist.
5. Prologue uses logging to print msgs.
6. Request and Response encapsulation increase overhead.
   Also lot of implementation defects will make it slow. I will improve its speed when I have free time.

[lihf8515]

Ok! Thank you very much!

…

---------- 该邮件从移动设备发送

--------------原始邮件-------------- 发件人："flywind "<notifications@github.com&gt;; 发送时间：2020年7月1日(星期三) 下午5:42 收件人："planety/prologue" <prologue@noreply.github.com&gt;; 抄送："lihaifeng "<lihaifeng@wxm.com&gt;;"Author "<author@noreply.github.com&gt;; 主题：Re: [planety/prologue] use "wrk" tool to test it for prologue. (#27) ----------------------------------- I haven't used valgrind to profile it. But I guess there may exists some aspects which caused its slowness: httpbeast uses nil HttpHeaders default, Prologue uses empty HttpHeaders default. Prologue uses middlewares which increases overhead. Prologue parses query parameters, form parameters and so on. It has startup, shutdown event, static file, to check whether exist. Prologue uses logging to print msgs. Request and Response encapsulation increase overhead. Also lot of implementation defects will make it slow. I will improve its speed when I have free time. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.