axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sam Ruby" <ru...@us.ibm.com>
Subject (update) Overall performance analyses of local GetStockQuote XXX
Date Tue, 20 Nov 2001 21:38:55 GMT
Environment summary:

   Win2K, Sun JDK 1.3.1, latest Axis from CVS, latest Xerces 2 from CVS.  OptimizeIt 4.0.2.

Results Summary (methods in which 2% or more of the time is spent):

   28.95% in SAXParser.parse (on the server side)
     includes 4.59% in schemaValidator.reset
     includes 3.84% in axis.DeserializationContext.startElement
     includes 2.01% in DTDValidator.reset
     includes 0.89% in DTDScannerImpl.reset

   26.82% in SAXParser.parse (on the client side)
     includes 4.13% in schemaValidator.reset
     includes 3.07% in axis.DeserializationContext.startElement
     includes 1.98% in DTDValidator.reset
     includes 0.90% in DTDScannerImpl.reset

   7.84% in SOAPPart.getAsString (called by LocalResponder.invoke; server=>client)

   7.51% in SOAPPart.getAsString (called by LocalSender.invoke; client=>server)

   6.5% in SimpleAuthenticationHandler.invoke

   3.41% in Java.io.BufferedReader.<init>

   3.37% in RPCElement.deserialize

Results Analysis:

   49% of the time was in the XML parser (14.5% initializing unused subsystems)
   14% of the time was in converting to XML
   10% was in Axis Deserialization (including replay)
    7% was in Authentication (note: Stock quote was almost unmeasureable)
    3% was in initializing the java buffered reader

   15% is all other processing combined

Previous analysis:
   http://marc.theaimsgroup.com/?l=axis-dev&m=100205460207401&w=2

Details can be found at:
   http://www.apache.org/~rubys/AxisXerces2ByMethod-011120.html
   http://www.apache.org/~rubys/AxisXerces2Nested-011120.html

- Sam Ruby


Mime
View raw message