測試下多線程和 gevent 在 socket 服務端的小包表現能力，測試的方法不太嚴謹，有點屬於自娛自樂，要是有問題之處，請大家噴之！
每個連接都特意堵塞了 0.5 秒鐘！
圖片 21.1 pic
在大批量 tcp 測試下，threading 的開銷越來越大，所以造成了在並發數加大的情況下，出現 threading 崩潰的情況！gevent 是 libevent 和協程的融合，一個線程裡面都可以跑超多的協程！ 利用 libevent 做 io 堵塞的調度，gevent 體系下，同一時間只有一個任務在運行！
先來測試下多線程： 我們就不加線程池了
#!/usr/bin/env python # -*- coding: utf-8 -*- #xiaorui.ccimport sysimport socketimport timeimport threading #xiaorui.ccdef threads(port): s = socket.socket s.bind((\'0.0.0.0\', port)) s.listen(500) while True:cli, addr = s.acceptt = threading.Thread(target=handle_request, args=(cli, time.sleep))t.daemon = Truet.startdef handle_request(s, sleep): try:s.recv(1024)sleep(0.5) s.send(\'\'\'http/1.0 200 OK Hello World! \'\'\')s.shutdown(socket.SHUT_WR)print \'.\', except Exception, ex:print ex finally:sys.stdout.flushs.closeif __name__ == \'__main__\': threads(4444)
用 threading 跑 socket，每個連接堵塞的時間是 0.5
time ab -n 10000 -c 500 http://127.0.0.1:4444/This is ApacheBench, Version 2.3 <$Revision: 655654 $>Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/Licensed to The Apache Software Foundation, http://www.apache.org/Benchmarking 127.0.0.1 (be patient)Completed 1000 requestsCompleted 2000 requestsCompleted 3000 requestsCompleted 4000 requestsCompleted 5000 requestsCompleted 6000 requestsCompleted 7000 requestsCompleted 8000 requestsCompleted 9000 requestsCompleted 10000 requestsFinished 10000 requestsServer Software:Server Hostname:127.0.0.1Server Port: 4444Document Path: /Document Length:0 bytesConcurrency Level: 500Time taken for tests: 11.123 secondsComplete requests: 10000Failed requests:0Write errors: 0Total transferred: 470000 bytesHTML transferred: 0 bytesRequests per second: 899.01 [#/sec] (mean)Time per request: 556.166 [ms] (mean)Time per request: 1.112 [ms] (mean, across all concurrent requests)Transfer rate: 41.26 [Kbytes/sec] receivedConnection Times (ms) min mean[+/-sd] median maxConnect:0 33 177.0 0 1000Processing: 500 508 33.9 501 1132Waiting: 500 508 33.9 501 1132Total:500 541 201.8 501 2132Percentage of the requests served within a certain time (ms) 50% 501 66% 501 75% 502 80% 505 90% 522 95% 532 98% 1534 99% 1722 100% 2132 (longest request)real 0m11.145suser 0m0.210ssys 0m0.961s
圖片 21.2 pic
加到 800 的時候~
圖片 21.3 pic
gevent：
#xiaorui.ccimport sysimport socketimport timeimport geventfrom gevent import socketdef server(port): s = socket.socket s.bind((\'0.0.0.0\', port)) s.listen(500) while True:cli, addr = s.acceptgevent.spawn(handle_request, cli, gevent.sleep)def handle_request(s, sleep): try:data=s.recv(1024)sleep(0.5)s.send(\'\'\'http/1.0 200 OK Hello World! this is xiaorui.cc !!!\'\'\')print datarequest_string = "GET %s HTTP/1.1rnHost: %srnrnServer: xiaorui.ccn" %(\'index.html\', \'127.0.0.1\') s.send(request_string)s.shutdown(socket.SHUT_WR)print \'.\',『be killed』 except Exception, ex:print ex finally:s.closeif __name__ == \'__main__\': server(7777)
gevent 跑 socket 服務：
並發數值是 500 的時候！
time ab -n 10000 -c 500 http://127.0.0.1:7777/This is ApacheBench, Version 2.3 <$Revision: 655654 $>Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/Licensed to The Apache Software Foundation, http://www.apache.org/Benchmarking 127.0.0.1 (be patient)Completed 1000 requestsCompleted 2000 requestsCompleted 3000 requestsCompleted 4000 requestsCompleted 5000 requestsCompleted 6000 requestsCompleted 7000 requestsCompleted 8000 requestsCompleted 9000 requestsCompleted 10000 requestsFinished 10000 requestsServer Software:Server Hostname:127.0.0.1Server Port: 7777Document Path: /Document Length:0 bytesConcurrency Level: 500Time taken for tests: 11.312 secondsComplete requests: 10000Failed requests:0Write errors: 0Total transferred: 20000 bytesHTML transferred: 0 bytesRequests per second: 884.04 [#/sec] (mean)Time per request: 565.584 [ms] (mean)Time per request: 1.131 [ms] (mean, across all concurrent requests)Transfer rate: 1.73 [Kbytes/sec] receivedConnection Times (ms) min mean[+/-sd] median maxConnect:0 44 202.7 0 1001Processing: 500 513 10.1 511 707Waiting: 500 513 10.1 511 707Total:500 557 204.1 512 1525Percentage of the requests served within a certain time (ms) 50% 512 66% 515 75% 517 80% 519 90% 531 95% 552 98% 1521 99% 1523 100% 1525 (longest request)real 0m11.334suser 0m0.159ssys 0m0.730s
圖片 21.4 pic
服務端看到的信息都是正常的！
圖片 21.5 pic
並發是 1000 的時候：
time ab -n 10000 -c 1000 http://127.0.0.1:7777/This is ApacheBench, Version 2.3 <$Revision: 655654 $>Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/Licensed to The Apache Software Foundation, http://www.apache.org/Benchmarking 127.0.0.1 (be patient)Completed 1000 requestsCompleted 2000 requestsCompleted 3000 requestsCompleted 4000 requestsCompleted 5000 requestsCompleted 6000 requestsCompleted 7000 requestsCompleted 8000 requestsCompleted 9000 requestsCompleted 10000 requestsFinished 10000 requestsServer Software:Server Hostname:127.0.0.1Server Port: 7777Document Path: /Document Length:0 bytesConcurrency Level: 1000Time taken for tests: 7.406 secondsComplete requests: 10000Failed requests:0Write errors: 0Total transferred: 20000 bytesHTML transferred: 0 bytesRequests per second: 1350.22 [#/sec] (mean)Time per request: 740.623 [ms] (mean)Time per request: 0.741 [ms] (mean, across all concurrent requests)Transfer rate: 2.64 [Kbytes/sec] receivedConnection Times (ms) min mean[+/-sd] median maxConnect:0 175 491.7 0 3000Processing: 500 520 17.7 515 707Waiting: 500 520 17.7 515 707Total:500 695 492.5 517 3521Percentage of the requests served within a certain time (ms) 50% 517 66% 523 75% 538 80% 569 90% 1515 95% 1530 98% 1539 99% 3514 100% 3521 (longest request)real 0m7.428suser 0m0.208ssys 0m0.741s
當並發到 1500 的時候：
time ab -n 10000 -c 1500 http://127.0.0.1:7777/This is ApacheBench, Version 2.3 <$Revision: 655654 $>Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/Licensed to The Apache Software Foundation, http://www.apache.org/Benchmarking 127.0.0.1 (be patient)Completed 1000 requestsCompleted 2000 requestsCompleted 3000 requestsCompleted 4000 requestsCompleted 5000 requestsCompleted 6000 requestsCompleted 7000 requestsCompleted 8000 requestsCompleted 9000 requestsCompleted 10000 requestsFinished 10000 requestsServer Software:Server Hostname:127.0.0.1Server Port: 7777Document Path: /Document Length:0 bytesConcurrency Level: 1500Time taken for tests: 5.290 secondsComplete requests: 10000Failed requests:0Write errors: 0Total transferred: 20000 bytesHTML transferred: 0 bytesRequests per second: 1890.27 [#/sec] (mean)Time per request: 793.536 [ms] (mean)Time per request: 0.529 [ms] (mean, across all concurrent requests)Transfer rate: 3.69 [Kbytes/sec] receivedConnection Times (ms) min mean[+/-sd] median maxConnect:0 214 404.9 1 1003Processing: 500 522 23.0 514 716Waiting: 500 522 23.0 514 716Total:500 736 406.7 520 1712Percentage of the requests served within a certain time (ms) 50% 520 66% 558 75% 602 80% 1506 90% 1526 95% 1531 98% 1535 99% 1548 100% 1712 (longest request)real 0m5.313suser 0m0.275ssys 0m0.763s
出現了少量的報錯：
圖片 21.6 pic
gevent 可以加個隊列，來限制協程的數目，但是數目限制了，雖然穩定了，但是並發數上不去。
from gevent.pool import Poolpool = Pool(N)
這裡測試有點簡單，雖然已經安排了連接的堵塞，但是畢竟不符合業務。 有時間把後端的任務改成才 mongodb 取數據 ！
本文出自 「峰雲，就她了。」 博客，謝絕轉載！