You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mina.apache.org by "Serge Baranov (JIRA)" <ji...@apache.org> on 2009/02/15 05:15:59 UTC

[jira] Issue Comment Edited: (DIRMINA-609) Big performance drop in 2.0M2 CumulativeProtocolDecoder for large messages

    [ https://issues.apache.org/jira/browse/DIRMINA-609?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12673566#action_12673566 ] 

crazycoder edited comment on DIRMINA-609 at 2/14/09 8:14 PM:
----------------------------------------------------------------

RC1 trunk has a regression which breaks this test again: DIRMINA-663 . I've verified that it works fine with M3, but not with RC1 or M4.

      was (Author: crazycoder):
    RC1 trunk has a regression which breaks this test again: DIRMINA-663 . I've verified that it works fine with M3, but not with RC1.
  
> Big performance drop in 2.0M2 CumulativeProtocolDecoder for large messages
> --------------------------------------------------------------------------
>
>                 Key: DIRMINA-609
>                 URL: https://issues.apache.org/jira/browse/DIRMINA-609
>             Project: MINA
>          Issue Type: Bug
>          Components: Filter
>    Affects Versions: 2.0.0-M2
>         Environment: Mac OS X 10.5 (Intel), Java 1.5
>            Reporter: Matthew Phillips
>            Assignee: Julien Vermillard
>             Fix For: 2.0.0-M3
>
>         Attachments: mina_test_1.1.zip, mina_test_2.0.zip
>
>
> Porting from MINA 1.1 to 2.0M2, one of my tests pinpointed a huge drop in the performance of a CumulativeProtocolDecoder-based codec when transmitting large (>2MB) messages. Under MINA 1.1.x, the transfer was effectively immediate, under 2.0M2 the system seems to get bogged down and calls the doDecode () method in 1-second intervals, with slowly-increasing amounts of data. The system is fast quad-core system, lightly unloaded. Almost no CPU usage is recorded when running either test.
> The two attached ZIP files have test cases against 1.1 and 2.0M2 that reproduce this. Their trace outputs follow: the MINA 1.1 case terminates in under a second, the 2.0 case times out and fails after 5 seconds.
> MINA 1.1
> ----------------------------------------------------------------------
> time       = 14:04:00.934
> remaining  = 1024
> limit      = 1024
> capacity   = 1024
> time       = 14:04:00.946
> remaining  = 3072
> limit      = 3072
> capacity   = 4194304
> time       = 14:04:00.947
> remaining  = 7168
> limit      = 7168
> capacity   = 4194304
> time       = 14:04:00.947
> remaining  = 15360
> limit      = 15360
> capacity   = 4194304
> time       = 14:04:00.947
> remaining  = 31744
> limit      = 31744
> capacity   = 4194304
> time       = 14:04:00.947
> remaining  = 64512
> limit      = 64512
> capacity   = 4194304
> time       = 14:04:00.948
> remaining  = 130048
> limit      = 130048
> capacity   = 4194304
> time       = 14:04:00.948
> remaining  = 261120
> limit      = 261120
> capacity   = 4194304
> time       = 14:04:00.949
> remaining  = 523264
> limit      = 523264
> capacity   = 4194304
> time       = 14:04:00.951
> remaining  = 1047552
> limit      = 1047552
> capacity   = 4194304
> time       = 14:04:00.953
> remaining  = 1571840
> limit      = 1571840
> capacity   = 4194304
> time       = 14:04:00.954
> remaining  = 2096128
> limit      = 2096128
> capacity   = 4194304
> time       = 14:04:00.955
> remaining  = 2097156
> limit      = 2097156
> capacity   = 4194304
> frame size = 2097152
> 2.0
> ----------------------------------------------------------------------
> time       = 14:01:59.154
> remaining  = 2048
> limit      = 2048
> capacity   = 2048
> time       = 14:01:59.175
> remaining  = 6144
> limit      = 6144
> capacity   = 2097156
> time       = 14:01:59.176
> remaining  = 14336
> limit      = 14336
> capacity   = 2097156
> time       = 14:01:59.176
> remaining  = 30720
> limit      = 30720
> capacity   = 2097156
> time       = 14:01:59.176
> remaining  = 63488
> limit      = 63488
> capacity   = 2097156
> time       = 14:01:59.177
> remaining  = 129024
> limit      = 129024
> capacity   = 2097156
> time       = 14:01:59.177
> remaining  = 194560
> limit      = 194560
> capacity   = 2097156
> time       = 14:01:59.177
> remaining  = 196608
> limit      = 196608
> capacity   = 2097156
> time       = 14:02:00.153
> remaining  = 262144
> limit      = 262144
> capacity   = 2097156
> time       = 14:02:00.153
> remaining  = 294912
> limit      = 294912
> capacity   = 2097156
> time       = 14:02:01.153
> remaining  = 360448
> limit      = 360448
> capacity   = 2097156
> time       = 14:02:01.154
> remaining  = 393216
> limit      = 393216
> capacity   = 2097156
> time       = 14:02:02.154
> remaining  = 458752
> limit      = 458752
> capacity   = 2097156
> time       = 14:02:02.154
> remaining  = 491520
> limit      = 491520
> capacity   = 2097156
> time       = 14:02:03.154
> remaining  = 557056
> limit      = 557056
> capacity   = 2097156
> time       = 14:02:03.155
> remaining  = 589824
> limit      = 589824
> capacity   = 2097156
> time       = 14:02:04.154
> remaining  = 655360
> limit      = 655360
> capacity   = 2097156
> time       = 14:02:04.155
> remaining  = 688128
> limit      = 688128
> capacity   = 2097156

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.