Thank you for your benchmarking description. I have just tested it too.
My results are something around 330 req/s which is superior to previous results. Great ! The only problem I have encountered is that after something around 2000 requests the whole application stalls and locks and do not respond anymore. It seems like some ugly dead lock or something similar. I have tried it with postgres instead of derby with similar "dead-lock" result. Daniel On Mon, May 4, 2009 at 11:16 PM, David Pollak <feeder.of.the.be...@gmail.com > wrote: > Daniel, > > I did a little measurement of Lift and did some tuning. On my Core 920 > machine with 12GB of RAM running Ubuntu 9.04 and the JDK 1.6.0_13 in 64 bit > mode with Tomcat 6.0.18 (I ran into some bugs in Jetty while measuring with > ab). My baseline running (both locally and across the network) > > ab -c 10 -n 20000 http://yak.local:8080/user_mgt/login > > I was seeing about 1,500 pages/second. I used YourKit to do some profiling > and improved performance on my box to about 2,200 pages/second. Here are > the steps I went through. I created a new project: > > d...@yak:~/benchmark$ mvn archetype:create -U > -DarchetypeGroupId=net.liftweb -DarchetypeArtifactId=lift-archetype-basic > -DarchetypeVersion=1.1-SNAPSHOT -DremoteRepositories= > http://scala-tools.org/repo-snapshots -DgroupId=com.liftcode > -DartifactId=benchmark > > I cd'ed into benchmark and typed "mvn clean install" I copied the > "target/benchmark-1.0-SNAPSHOT" directory into Tomcat's webapps directory > and renamed the directory "ROOT". I did export > JAVA_OPTS="-Drun.mode=production" to make sure Lift runs in production > mode. I started Tomcat with ./bin/startup.sh > > Then I ran: > > d...@yak:~/tmp$ ab -c 10 -n 20000 http://localhost:8080/user_mgt/login > > I did this and discarded the results because this will "warm up" the > HotSpot compiler. At the end of this run, the JVM should have optimized > most of the Lift-related classes. Next, I re-ran the command: > d...@yak:~/tmp$ ab -c 10 -n 20000 http://localhost:8080/user_mgt/login > This is ApacheBench, Version 2.3 <$Revision: 655654 $> > Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/ > Licensed to The Apache Software Foundation, http://www.apache.org/ > > Benchmarking localhost (be patient) > Completed 2000 requests > Completed 4000 requests > Completed 6000 requests > Completed 8000 requests > Completed 10000 requests > Completed 12000 requests > Completed 14000 requests > Completed 16000 requests > Completed 18000 requests > Completed 20000 requests > Finished 20000 requests > > > Server Software: Apache-Coyote/1.1 > Server Hostname: localhost > Server Port: 8080 > > Document Path: /user_mgt/login > Document Length: 3768 bytes > > Concurrency Level: 10 > Time taken for tests: 9.156 seconds > Complete requests: 20000 > Failed requests: 0 > Write errors: 0 > Total transferred: 83620000 bytes > HTML transferred: 75360000 bytes > Requests per second: 2184.33 [#/sec] (mean) > Time per request: 4.578 [ms] (mean) > Time per request: 0.458 [ms] (mean, across all concurrent requests) > Transfer rate: 8918.62 [Kbytes/sec] received > > Connection Times (ms) > min mean[+/-sd] median max > Connect: 0 1 0.9 0 8 > Processing: 1 4 18.5 3 788 > Waiting: 0 4 18.4 2 788 > Total: 1 5 18.5 3 788 > WARNING: The median and mean for the initial connection time are not within > a normal deviation > These results are probably not that reliable. > > Percentage of the requests served within a certain time (ms) > 50% 3 > 66% 4 > 75% 4 > 80% 5 > 90% 6 > 95% 8 > 98% 11 > 99% 16 > 100% 788 (longest request) > d...@yak:~/tmp$ > > I also ran the command remotely over a gigabit LAN: > d...@horse:~$ ab -c 10 -n 20000 http://yak.local:8080/user_mgt/login > 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 yak.local (be patient) > Completed 2000 requests > Completed 4000 requests > Completed 6000 requests > Completed 8000 requests > Completed 10000 requests > Completed 12000 requests > Completed 14000 requests > Completed 16000 requests > Completed 18000 requests > Finished 20000 requests > > > Server Software: Apache-Coyote/1.1 > Server Hostname: yak.local > Server Port: 8080 > > Document Path: /user_mgt/login > Document Length: 3768 bytes > > Concurrency Level: 10 > Time taken for tests: 10.305721 seconds > Complete requests: 20000 > Failed requests: 0 > Write errors: 0 > Total transferred: 83620000 bytes > HTML transferred: 75360000 bytes > Requests per second: 1940.67 [#/sec] (mean) > Time per request: 5.153 [ms] (mean) > Time per request: 0.515 [ms] (mean, across all concurrent requests) > Transfer rate: 7923.75 [Kbytes/sec] received > > Connection Times (ms) > min mean[+/-sd] median max > Connect: 0 0 0.6 0 4 > Processing: 1 4 25.9 3 840 > Waiting: 1 3 25.6 2 816 > Total: 1 4 25.9 3 840 > > Percentage of the requests served within a certain time (ms) > 50% 3 > 66% 4 > 75% 4 > 80% 4 > 90% 6 > 95% 8 > 98% 11 > 99% 17 > 100% 840 (longest request) > > > I'm currently setting up a test machine with a 2 core AMD processor, 1GB of > RAM and Ubuntu. I'll re-run the tests on this box and let you know. > > Thanks, > > David > > > > > > > On Sat, May 2, 2009 at 1:32 PM, Dunsun <dun...@gmail.com> wrote: > >> >> Hi, >> >> I'm very new to lift and scala. >> >> I have made some Grails intranet applications but sadly Grails or >> better say GSP rendering part which is completely written in Groovy is >> a very slow beast (even in comparison with Rails or Django). >> >> For my future big project (heavy loaded site) I need something much >> faster. >> >> So I have made some quick and dirty Lift benchmarks using ApacheBench >> + Tomcat, GlassFish (from netbeans) and even Jetty. >> My machine - AMD X2 @ 2500Mhz, 4 GB RAM >> >> LIFT BASIC SAMPLE: >> ab -c 10 -n 3000 >> http://localhost:8080/liftbasic-1.0-SNAPSHOT/user_mgt/login >> 50 req/s >> >> LIFT BLANK SAMPLE: >> ab -c 10 -n 3000 http://localhost:8080/liftblank-1.0/ >> 120 req/s >> >> Obtained numbers were similar for all 3 servers. >> >> Using Stripes and rendering similar simple pages I'm getting much much >> higher scores (600 req/s). >> Using Grails I'm getting little worse results but I would say it is >> same league. >> I would expect much better performance from Lift which is written in >> scala. >> >> Why am I getting these low numbers ? >> Am I doing something wrong ? >> Any suggestion is very welcome. >> >> regards >> Daniel >> >> >> > > > -- > Lift, the simply functional web framework http://liftweb.net > Beginning Scala http://www.apress.com/book/view/1430219890 > > Follow me: http://twitter.com/dpp > Git some: http://github.com/dpp > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Lift" group. To post to this group, send email to liftweb@googlegroups.com To unsubscribe from this group, send email to liftweb+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/liftweb?hl=en -~----------~----~----~----~------~----~------~--~---
