You are viewing a plain text version of this content. The canonical link for it is here.
Posted to soap-user@xml.apache.org by Paul McKernon <p....@aepona.com> on 2000/08/08 15:46:44 UTC
SOAP performance
I would say that there are a number of factors that influenced my testing e.g.
SOAP has not been commercially refined whereas CORBA has
SOAP uses Java whereas CORBA is using C++
The payload of a request is small (a small object filled with text) which
creates a larger overhead because of the XML wrapping.
I can see many reasons why SOAP will be slower and although C++ implementation
and refinement will come - they are not present at this moment. I am not trying
to 'knock' SOAP (personally I like it better than CORBA) but simply trying to
find out which technology performs best at this moment in time for simple data
transfer.
I am looking for other people that have done similar tests to conform / rubbish
my findings.
Paul.
"Cabrera, Alan" wrote:
> Could it be that it's because you compared a reference implementation of
> SOAP with a production imeplmentation of CORBA?
>
> -----Original Message-----
> From: Paul McKernon [mailto:p.mckernon@aepona.com]
> Sent: Tuesday, August 08, 2000 9:19 AM
> To: soap-user@xml.apache.org
> Subject: Re: Address Book Error
>
> Has anyone in the group done any performance testing of SOAP vs CORBA?
>
> I have done some basic testing and discovered that CORBA is many times
> faster than SOAP ( a very simple application).
> I used Tomcat 3.1 / Xerces 1.1.3 / apache soap v 2.0 on Linux. Can anyone
> tell me if they found similar /
> conradicting results?
>
> Also - is soap useful for two way communications as is implemented in CORBA?
> Can I have two objects on two different
> machines that can communicate bidirectionally? Are there any tools out there
> that can facilitate this kind of setup?
>
> Many Thanks for any help,
> Paul
Re: SOAP performance
Posted by Dennis Sosnoski <dm...@sosnoski.com>.
I doubt that the Java implementation of SOAP is a major contributor to the speed
difference. If your main concern is performance you're always going to be much
better off using CORBA than SOAP because CORBA uses a binary wire format for the
data transfer, as opposed to the verbosity of XML text. You can reduce the
impact of this with compression of the XML, but that just increases the already
higher processor load of SOAP (generating and parsing all that XML text, as
opposed to just writing/reading binary).
SOAP is great for easy interoperability (once the implementations stabalize!)
and flexibility - I'm thinking of some applications for running SOAP documents
through XSLT, for instance. Performance will definitely improve substantially,
but it's unlikely to ever be a strong point.
- Dennis
Paul McKernon wrote:
>
> I would say that there are a number of factors that influenced my testing e.g.
>
> SOAP has not been commercially refined whereas CORBA has
> SOAP uses Java whereas CORBA is using C++
> The payload of a request is small (a small object filled with text) which
> creates a larger overhead because of the XML wrapping.
>
> I can see many reasons why SOAP will be slower and although C++ implementation
> and refinement will come - they are not present at this moment. I am not trying
> to 'knock' SOAP (personally I like it better than CORBA) but simply trying to
> find out which technology performs best at this moment in time for simple data
> transfer.
>
> I am looking for other people that have done similar tests to conform / rubbish
> my findings.
>
> Paul.
>
> "Cabrera, Alan" wrote:
>
> > Could it be that it's because you compared a reference implementation of
> > SOAP with a production imeplmentation of CORBA?
> >
> > -----Original Message-----
> > From: Paul McKernon [mailto:p.mckernon@aepona.com]
> > Sent: Tuesday, August 08, 2000 9:19 AM
> > To: soap-user@xml.apache.org
> > Subject: Re: Address Book Error
> >
> > Has anyone in the group done any performance testing of SOAP vs CORBA?
> >
> > I have done some basic testing and discovered that CORBA is many times
> > faster than SOAP ( a very simple application).
> > I used Tomcat 3.1 / Xerces 1.1.3 / apache soap v 2.0 on Linux. Can anyone
> > tell me if they found similar /
> > conradicting results?
> >
> > Also - is soap useful for two way communications as is implemented in CORBA?
> > Can I have two objects on two different
> > machines that can communicate bidirectionally? Are there any tools out there
> > that can facilitate this kind of setup?
> >
> > Many Thanks for any help,
> > Paul
Re: SOAP performance
Posted by Dennis Sosnoski <dm...@sosnoski.com>.
I doubt that the Java implementation of SOAP is a major contributor to the speed
difference. If your main concern is performance you're always going to be much
better off using CORBA than SOAP because CORBA uses a binary wire format for the
data transfer, as opposed to the verbosity of XML text. You can reduce the
impact of this with compression of the XML, but that just increases the already
higher processor load of SOAP (generating and parsing all that XML text, as
opposed to just writing/reading binary).
SOAP is great for easy interoperability (once the implementations stabalize!)
and flexibility - I'm thinking of some applications for running SOAP documents
through XSLT, for instance. Performance will definitely improve substantially,
but it's unlikely to ever be a strong point.
- Dennis
Paul McKernon wrote:
>
> I would say that there are a number of factors that influenced my testing e.g.
>
> SOAP has not been commercially refined whereas CORBA has
> SOAP uses Java whereas CORBA is using C++
> The payload of a request is small (a small object filled with text) which
> creates a larger overhead because of the XML wrapping.
>
> I can see many reasons why SOAP will be slower and although C++ implementation
> and refinement will come - they are not present at this moment. I am not trying
> to 'knock' SOAP (personally I like it better than CORBA) but simply trying to
> find out which technology performs best at this moment in time for simple data
> transfer.
>
> I am looking for other people that have done similar tests to conform / rubbish
> my findings.
>
> Paul.
>
> "Cabrera, Alan" wrote:
>
> > Could it be that it's because you compared a reference implementation of
> > SOAP with a production imeplmentation of CORBA?
> >
> > -----Original Message-----
> > From: Paul McKernon [mailto:p.mckernon@aepona.com]
> > Sent: Tuesday, August 08, 2000 9:19 AM
> > To: soap-user@xml.apache.org
> > Subject: Re: Address Book Error
> >
> > Has anyone in the group done any performance testing of SOAP vs CORBA?
> >
> > I have done some basic testing and discovered that CORBA is many times
> > faster than SOAP ( a very simple application).
> > I used Tomcat 3.1 / Xerces 1.1.3 / apache soap v 2.0 on Linux. Can anyone
> > tell me if they found similar /
> > conradicting results?
> >
> > Also - is soap useful for two way communications as is implemented in CORBA?
> > Can I have two objects on two different
> > machines that can communicate bidirectionally? Are there any tools out there
> > that can facilitate this kind of setup?
> >
> > Many Thanks for any help,
> > Paul