axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From bugzi...@apache.org
Subject DO NOT REPLY [Bug 11539] New: - Binary data transfer performance appears suboptimal
Date Wed, 07 Aug 2002 20:33:25 GMT
DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG 
RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT
<http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11539>.
ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND 
INSERTED IN THE BUG DATABASE.

http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11539

Binary data transfer performance appears suboptimal

           Summary: Binary data transfer performance appears suboptimal
           Product: Axis
           Version: beta-3
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: Normal
          Priority: Other
         Component: Basic Architecture
        AssignedTo: axis-dev@xml.apache.org
        ReportedBy: rkimbrell@northropgrumman.com


See also bug #11532.

See the attachment.  This test (also described in bug #11532) was to estimate
the ability of the Axis web server to provide large blocks of data.  A simple
client and server were prepared.  These were based on the client/server
described in the August 2002 Linux Magazine article (thus the internal naming of
the classes, etc.).  The server provides a block of data upon request of the
client.  The block size could be 4096, 8192, ...., 1048576 bytes: 4096*2^0,
4096*2^1, 4096*2^2, 4096*2^3, ..., 4096*2^11.  The block was filled with a byte
value corresponding to the power of two.  The results of the test are detailed
in the attachment, an MS Excel spreadsheet (sorry, but that's the environment I
work in).  

As expected the transmission time increases with block size.  The optimal block
size was in the neighborhood of 131K.  The transmission rate was about 25K
Bytes/sec.  This is a pronounced peak.  It falls off after that.  At a block
size of about 1 Mbyte, the transmission rate is less than 6 Kbytes/sec.  

I expected to see the transmission rate increase - after all, fixed overhead
should become less of a factor as block size increses.  However, I would not
expect transmission rates to fall off so rapidly above 131K.  I can see some of
this as a factor of buffering, but this seemes excessive.

It could be that the test or my ideas are faulty here.  If so, I'd appreciate
comments.

Thanks,
Roy

Mime
View raw message