On my Ubuntu 8.04 64 bit desktop, I created an empty Rails project and from another machine, I used ab -n 10 http://210.77.27.169:3000/ to test the performance:
On my Ubuntu 8.04 64 bit desktop, I created an empty Rails project and from another machine, I used ab -n 10http://210.77.27.169:3000/ to test the performance:
That's interesting, I didn't get anything like that difference locally. in fact rails 2.3.3 was faster.
Fred
When you did this "test" did you run the tests multiple times and take the average? Was this something that was reproducible?
Were they both running in production mode?
Colin
Yes, I ran it several times. Both on a Windows machine of my colleague and my Linux box. In fact, it is my colleague complaining about the low efficiency when testing it locally after I upgrade rails from 2.2.2 to 2.3.3 that drove me to do this test.
No, they all ran in development mode. In our production machines, I did not notice such huge difference in performance. If it is because of the development mode, I wonder: 1. What slows Rails 2.3.3 down in development? 2. How to overcome it? We need to test our app in our development machine, this slows down our development significantly and makes the developers very frustrated when they need to wait for each request.
Did you visit your app by localhost or by your IP address? Visiting by localhost is quite fast.
No, they all ran in development mode. In our production machines, I did not notice such huge difference in performance. If it is because of the development mode, I wonder: 1. What slows Rails 2.3.3 down in development? 2. How to overcome it? We need to test our app in our development machine, this slows down our development significantly and makes the developers very frustrated when they need to wait for each request.
Does your other message implies that the app responds quickly when accessing it from localhost ? (I was. if that is the discriminating factor then it is very curious). Also it was my impression that webrick/mongrel only pass the request to rails if there isn't a file in public that matches (ie hitting / just returns index.html with basically no rails interaction)
Fred
Now I create two rails apps one by 2.2.2, the other by 2.3.3, each created with users scaffolding. Then I access /users/ to touch rails. Each test is performed several times. Accessing from local machine, the performance seem to be no much difference; it's both about 40ms per request. However, accessing from a remote server, there's a dramatic difference: ================Rails 2.3.3================ hoolai@db:~$ ab -n 5 http://210.77.26.176:3000/users/ This is ApacheBench, Version 2.0.40-dev <$Revision: 1.146 $> apache-2.0 Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/ Copyright 2006 The Apache Software Foundation, http://www.apache.org/
Benchmarking 210.77.26.176 (be patient).....done
Server Software: WEBrick/1.3.1 Server Hostname: 210.77.26.176 Server Port: 3000
Document Path: /users/ Document Length: 1301 bytes
Concurrency Level: 1 Time taken for tests: 50.448329 seconds Complete requests: 5 Failed requests: 0 Write errors: 0 Total transferred: 9200 bytes HTML transferred: 6505 bytes Requests per second: 0.10 [#/sec] (mean) Time per request: 10089.666 [ms] (mean) Time per request: 10089.666 [ms] (mean, across all concurrent requests) Transfer rate: 0.16 [Kbytes/sec] received
Connection Times (ms) min mean[+/-sd] median max Connect: 2 5 4.7 6 12 Processing: 10061 10083 22.3 10080 10121 Waiting: 10060 10082 21.9 10078 10119 Total: 10063 10089 26.3 10085 10133
Percentage of the requests served within a certain time (ms) 50% 10082 66% 10089 75% 10089 80% 10133 90% 10133 95% 10133 98% 10133 99% 10133 100% 10133 (longest request) =====================Rails 2.2.2================ hoolai@db:~$ ab -n 5 http://210.77.26.176:3000/users/ This is ApacheBench, Version 2.0.40-dev <$Revision: 1.146 $> apache-2.0 Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/ Copyright 2006 The Apache Software Foundation, http://www.apache.org/
Benchmarking 210.77.26.176 (be patient).....done
Server Software: WEBrick/1.3.1 Server Hostname: 210.77.26.176 Server Port: 3000
Document Path: /users/ Document Length: 1298 bytes
Concurrency Level: 1 Time taken for tests: 0.307687 seconds Complete requests: 5 Failed requests: 0 Write errors: 0 Total transferred: 9140 bytes HTML transferred: 6490 bytes Requests per second: 16.25 [#/sec] (mean) Time per request: 61.537 [ms] (mean) Time per request: 61.537 [ms] (mean, across all concurrent requests) Transfer rate: 26.00 [Kbytes/sec] received
Connection Times (ms) min mean[+/-sd] median max Connect: 2 3 2.2 4 6 Processing: 35 57 46.3 37 140 Waiting: 34 56 46.4 36 139 Total: 37 61 45.9 43 143
Percentage of the requests served within a certain time (ms) 50% 43 66% 43 75% 43 80% 143 90% 143 95% 143 98% 143 99% 143 100% 143 (longest request)
When I looked at the console output when Rails 2.3.3 was processing request, it seemed that each request was responded quite fast. I don't understand why it took so long for the client to receive the response. Is it because of rack?
Processing UsersController#index (for 222.35.3.195 at 2009-09-06 09:54:58) [GET] User Load (0.2ms) SELECT * FROM `users` Rendering template within layouts/users Rendering users/index User Columns (1.6ms) SHOW FIELDS FROM `users` Completed in 21ms (View: 7, DB: 2) | 200 OK [http://210.77.26.176/ users/]
