[
https://issues.apache.org/jira/browse/LOG4J2-2644?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16880031#comment-16880031
]
Ralph Goers commented on LOG4J2-2644:
-------------------------------------
Performance results comparing before any changes and after all changes.
{code:java}
Release 2.x Before Changes 4 Threads
Benchmark Mode Cnt Score Error
Units
FileAppenderBenchmark.julFile thrpt 20 30536.101 ±
858.591 ops/s
FileAppenderBenchmark.log4j1File thrpt 20 617306.350 ±
16010.513 ops/s
FileAppenderBenchmark.log4j2AsyncAppender thrpt 20 435017.227 ±
17070.693 ops/s
FileAppenderBenchmark.log4j2AsyncDisruptor thrpt 20 959897.397 ±
28078.800 ops/s
FileAppenderBenchmark.log4j2AsyncLogger thrpt 20 1129097.193 ±
38648.100 ops/s
FileAppenderBenchmark.log4j2File thrpt 20 1041400.912 ±
32562.135 ops/s
FileAppenderBenchmark.log4j2RAF thrpt 20 1112014.290 ±
22171.312 ops/s
FileAppenderBenchmark.logbackAsyncFile thrpt 20 468825.311 ±
20857.864 ops/s
FileAppenderBenchmark.logbackFile thrpt 20 978704.022 ±
23239.574 ops/s
Benchmark Mode Cnt Score
Error Units
FileAppenderWithLocationBenchmark.log4j1File thrpt 20 49509.572 ±
6261.952 ops/s
FileAppenderWithLocationBenchmark.log4j2File thrpt 20 156442.483 ±
6253.948 ops/s
FileAppenderWithLocationBenchmark.log4j2RAF thrpt 20 145652.582 ±
7432.223 ops/s
FileAppenderWithLocationBenchmark.logbackFile thrpt 20 144273.461 ±
6404.892 ops/s
Release 2.x After Changes 4 Threads
Benchmark Mode Cnt Score Error
Units
FileAppenderBenchmark.julFile thrpt 20 42467.810 ±
1769.847 ops/s
FileAppenderBenchmark.log4j1File thrpt 20 635168.969 ±
18965.265 ops/s
FileAppenderBenchmark.log4j2AsyncAppender thrpt 20 450376.886 ±
25937.515 ops/s
FileAppenderBenchmark.log4j2AsyncDisruptor thrpt 20 1012156.176 ±
23415.378 ops/s
FileAppenderBenchmark.log4j2AsyncLogger thrpt 20 1259536.386 ±
31958.779 ops/s
FileAppenderBenchmark.log4j2File thrpt 20 1166042.327 ±
24917.504 ops/s
FileAppenderBenchmark.log4j2RAF thrpt 20 1205182.808 ±
32388.393 ops/s
FileAppenderBenchmark.logbackAsyncFile thrpt 20 533475.105 ±
12298.555 ops/s
FileAppenderBenchmark.logbackFile thrpt 20 1180302.637 ±
16191.367 ops/s
Benchmark Mode Cnt Score
Error Units
FileAppenderWithLocationBenchmark.log4j1File thrpt 20 55333.318 ±
1056.748 ops/s
FileAppenderWithLocationBenchmark.log4j2File thrpt 20 193114.226 ±
3694.770 ops/s
FileAppenderWithLocationBenchmark.log4j2RAF thrpt 20 210400.326 ±
3427.059 ops/s
FileAppenderWithLocationBenchmark.logbackFile thrpt 20 129510.641 ±
2934.328 ops/s
Release 3.x Before Changes 4 Threads
Benchmark Mode Cnt Score Error
Units
FileAppenderBenchmark.julFile thrpt 20 30124.191 ± 5789.724
ops/s
FileAppenderBenchmark.log4j1File thrpt 20 376516.283 ± 27072.069
ops/s
FileAppenderBenchmark.log4j2AsyncAppender thrpt 20 294996.863 ± 15507.474
ops/s
FileAppenderBenchmark.log4j2AsyncDisruptor thrpt 20 573890.697 ± 26375.565
ops/s
FileAppenderBenchmark.log4j2AsyncLogger thrpt 20 520976.275 ± 27280.963
ops/s
FileAppenderBenchmark.log4j2File thrpt 20 404600.346 ± 32764.188
ops/s
FileAppenderBenchmark.log4j2RAF thrpt 20 588507.044 ± 15255.207
ops/s
FileAppenderBenchmark.logbackAsyncFile thrpt 20 245658.860 ± 10594.429
ops/s
FileAppenderBenchmark.logbackFile thrpt 20 506914.909 ± 6268.274
ops/s
Benchmark Mode Cnt Score
Error Units
FileAppenderWithLocationBenchmark.log4j1File thrpt 20 47113.312 ±
604.509 ops/s
FileAppenderWithLocationBenchmark.log4j2File thrpt 20 128269.166 ±
3856.169 ops/s
FileAppenderWithLocationBenchmark.log4j2RAF thrpt 20 128673.823 ±
2625.317 ops/s
FileAppenderWithLocationBenchmark.logbackFile thrpt 20 123472.627 ±
2211.028 ops/s
Release 3.x After Changes 4 Threads
Benchmark Mode Cnt Score Error
Units
FileAppenderBenchmark.julFile thrpt 20 33326.135 ±
918.762 ops/s
FileAppenderBenchmark.log4j1File thrpt 20 594480.764 ±
15160.389 ops/s
FileAppenderBenchmark.log4j2AsyncAppender thrpt 20 319499.481 ±
5545.647 ops/s
FileAppenderBenchmark.log4j2AsyncDisruptor thrpt 20 972707.271 ±
33654.138 ops/s
FileAppenderBenchmark.log4j2AsyncLogger thrpt 20 1038696.405 ±
62221.518 ops/s
FileAppenderBenchmark.log4j2File thrpt 20 1079429.488 ±
28685.159 ops/s
FileAppenderBenchmark.log4j2RAF thrpt 20 1113956.149 ±
21609.459 ops/s
FileAppenderBenchmark.logbackAsyncFile thrpt 20 438444.717 ±
9520.599 ops/s
FileAppenderBenchmark.logbackFile thrpt 20 984278.947 ±
28560.305 ops/s
Benchmark Mode Cnt Score
Error Units
FileAppenderWithLocationBenchmark.log4j1File thrpt 20 55437.605 ±
1154.674 ops/s
FileAppenderWithLocationBenchmark.log4j2File thrpt 20 176225.467 ±
7695.072 ops/s
FileAppenderWithLocationBenchmark.log4j2FluentFile thrpt 20 507453.923 ±
12238.743 ops/s
FileAppenderWithLocationBenchmark.log4j2RAF thrpt 20 166023.981 ±
7193.730 ops/s
FileAppenderWithLocationBenchmark.logbackFile thrpt 20 147522.065 ±
3045.869 ops/s
{code}
> Logging with location information is considerably slower than logging with
> location information in java.util.logging
> --------------------------------------------------------------------------------------------------------------------
>
> Key: LOG4J2-2644
> URL: https://issues.apache.org/jira/browse/LOG4J2-2644
> Project: Log4j 2
> Issue Type: Improvement
> Affects Versions: 2.11.2
> Reporter: Marco Herrn
> Priority: Minor
> Time Spent: 2h 40m
> Remaining Estimate: 0h
>
> according to https://logging.apache.org/log4j/2.x/performance.html logging
> with location information should be roughly the same performance for JUL
> and Log4j2.
> However I made a few comparisons between Log4j2 and JUL and in this case
> Log4j2 is much faster than JUL when omitting location information, but is
> considerably slower when logging with location information.
> Those are the results:
> JUL without location information : 32.162s
> JUL with location information : 21.682s
> Log4j2 without location information: 4.168s
> Log4j2 with location information : 59.000s
> In each case I logged 1500000 simple log statements (only a fixed string)
> and compared the timestamp of the first and the last generated log
> statement. Above you see the time spent between the first and the last log
> statement.
> I did these tests several times and the results are equal each time.
> In all cases I logged to a RollingFileAppender (or JULs equivalent).
> I assume that JUL with location information is faster than JUL without
> location information is because for JUL with location information I used a
> custom Formatter that doesn't support any configuration whereas I was
> using a SimpleFormatter with a specified formatstring for JUL without log
> information.
> It should be noted that I didn't use the Log4j2 API, but instead used the
> JUL logging API and used the log4j-jul-bridge to actually use Log4j2
> instead of the JUL implementation!
> I want to pay special attention to the difference when logging with
> location information, since I am puzzled that Log4j2 is that much slower
> than JUL in that case.
> The example code used for testing and the corresponding logging
> configurations can be found at [my example github
> project|http://github.com/hupfdule/LoggingPerformanceTest].
> I don't think there is anything special. Since when logging without
> location information Log4j2 is much faster than JUL there must be a
> difference between getting the location information. My above mentioned
> custom formatter uses the methods java.util.LogRecord#getSourceClassName()
> and java.util.LogRecord#getSourceMethodName() for obtaining that
> information. I haven't looked into how Log4j2 does it, but it seems to be
> much less efficient.
> I have done the tests with Java 8, but it seems that running it under Java 11
> shows the same behaviour.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
