You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-user@axis.apache.org by Vy Ho <st...@drexel.edu> on 2004/11/16 17:12:15 UTC
Signal to noise ratio in Soap is very high
I look at a message send back from the server to a client in Axis, and
was very surprised that most of the text is not the data that I need to
send, but declaration and long namspaces. Is it possible to reduce
these down and still comply with Soap?
Thanks,
vh
Re: Signal to noise ratio in Soap is very high
Posted by Jim Murphy <jm...@mindreef.com>.
Can someone describe what the "slow part of web services is"? If you
think its the verbosity of XML I suspect your not measuring it against
alternatives. The fact is unless your message payload is very large the
size of the XML body matters little in your overall performance - over
broadband.
A great example of this is trying to optimize SOAP envelopes in the <
1500 byte range - where many SOAP envelopes fall. In this common case
the ethernet frame size is packed up to ~1500 when you don't use all
~1500 so "optimizing" from 1000 bytes to 500 just costs you. Its costs
you the time and complexity of the compression step.
When message sizes get larger it's still not the perf killer you may
think - until messages sizes exceed !MB - YMMV.
Jim Murphy
Mindreef, Inc.
salgado.pc@04web.com wrote:
> I read an article about Fast Web Services... but I think there was no
> new development since then.
>
> http://java.sun.com/developer/technicalArticles/WebServices/fastWS/
>
>
>
>>Thank you for the responses. I think it's possible to achieve
>>interoperability and small foot print. It's just Soap was designed
>>lousy in this particular aspect. When saying if speed matters, I think
>>most of the apps that has some kind of communication, speed does
>>matter. I think there should be a new standard/version to replace Soap
>>if this cannot be addressed.
>>
>>Paul Callahan wrote:
>>
>>
>>>I second that... And would just add that you might want to think about
>>>when your application is just too 'chatty' for SOAP...
>>>
>>>-pc
>>>
>>>On Tue, 16 Nov 2004 08:48:30 -0900, Elaine Nance
>>><el...@commerce.state.ak.us> wrote:
>>>
>>>
>>>
>>>>SOAP is not designed to create compact dataflows. It is designed for
>>>>interoperability and, because it is XML, extensibility.
>>>>IMHO, I would *not* use SOAP for purely internal processes if
>>>>speed or stream size matters (where faster/smaller) is better.
>>>>
>>>>In that case I would go strictly with a socket-level approach.
>>>>
>>>>Just my 2 small coins,
>>>>Elaine
>>>>
>>>>
>>>>
>>>>Vy Ho wrote:
>>>>
>>>>
>>>>
>>>>>I look at a message send back from the server to a client in Axis,
>>>>>and was very surprised that most of the text is not the data that I
>>>>>need to send, but declaration and long namspaces. Is it possible to
>>>>>reduce these down and still comply with Soap?
>>>>>
>>>>>Thanks,
>>>>>
>>>>>vh
>>>>>
>>>>>
>>>>>
>>>>
>>>><~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>> | Computers are useless. They can only give you answers.
>>>> | -- Pablo Picasso --
>>>><~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>>
>>>>
>>>>
>>>>
>
>
>
>
>
Re: Signal to noise ratio in Soap is very high
Posted by sa...@04web.com.
I read an article about Fast Web Services... but I think there was no
new development since then.
http://java.sun.com/developer/technicalArticles/WebServices/fastWS/
>
> Thank you for the responses. I think it's possible to achieve
> interoperability and small foot print. It's just Soap was designed
> lousy in this particular aspect. When saying if speed matters, I think
> most of the apps that has some kind of communication, speed does
> matter. I think there should be a new standard/version to replace Soap
> if this cannot be addressed.
>
> Paul Callahan wrote:
>
>>I second that... And would just add that you might want to think about
>> when your application is just too 'chatty' for SOAP...
>>
>>-pc
>>
>>On Tue, 16 Nov 2004 08:48:30 -0900, Elaine Nance
>><el...@commerce.state.ak.us> wrote:
>>
>>
>>>SOAP is not designed to create compact dataflows. It is designed for
>>> interoperability and, because it is XML, extensibility.
>>>IMHO, I would *not* use SOAP for purely internal processes if
>>>speed or stream size matters (where faster/smaller) is better.
>>>
>>>In that case I would go strictly with a socket-level approach.
>>>
>>>Just my 2 small coins,
>>>Elaine
>>>
>>>
>>>
>>>Vy Ho wrote:
>>>
>>>
>>>>I look at a message send back from the server to a client in Axis,
>>>> and was very surprised that most of the text is not the data that I
>>>> need to send, but declaration and long namspaces. Is it possible to
>>>> reduce these down and still comply with Soap?
>>>>
>>>>Thanks,
>>>>
>>>>vh
>>>>
>>>>
>>>>
>>><~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>> | Computers are useless. They can only give you answers.
>>> | -- Pablo Picasso --
>>><~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>
>>>
>>>
>>>
Re: Signal to noise ratio in Soap is very high
Posted by Vy Ho <st...@drexel.edu>.
Thank you for the responses. I think it's possible to achieve
interoperability and small foot print. It's just Soap was designed
lousy in this particular aspect. When saying if speed matters, I think
most of the apps that has some kind of communication, speed does
matter. I think there should be a new standard/version to replace Soap
if this cannot be addressed.
Paul Callahan wrote:
>I second that... And would just add that you might want to think
>about when your application is just too 'chatty' for SOAP...
>
>-pc
>
>On Tue, 16 Nov 2004 08:48:30 -0900, Elaine Nance
><el...@commerce.state.ak.us> wrote:
>
>
>>SOAP is not designed to create compact dataflows. It is designed
>>for interoperability and, because it is XML, extensibility.
>>IMHO, I would *not* use SOAP for purely internal processes if
>>speed or stream size matters (where faster/smaller) is better.
>>
>>In that case I would go strictly with a socket-level approach.
>>
>>Just my 2 small coins,
>>Elaine
>>
>>
>>
>>Vy Ho wrote:
>>
>>
>>>I look at a message send back from the server to a client in Axis, and
>>>was very surprised that most of the text is not the data that I need to
>>>send, but declaration and long namspaces. Is it possible to reduce
>>>these down and still comply with Soap?
>>>
>>>Thanks,
>>>
>>>vh
>>>
>>>
>>>
>><~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> | Computers are useless. They can only give you answers.
>> | -- Pablo Picasso --
>><~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>
>>
>>
>>
>>
Re: Signal to noise ratio in Soap is very high
Posted by Paul Callahan <ca...@gmail.com>.
I second that... And would just add that you might want to think
about when your application is just too 'chatty' for SOAP...
-pc
On Tue, 16 Nov 2004 08:48:30 -0900, Elaine Nance
<el...@commerce.state.ak.us> wrote:
> SOAP is not designed to create compact dataflows. It is designed
> for interoperability and, because it is XML, extensibility.
> IMHO, I would *not* use SOAP for purely internal processes if
> speed or stream size matters (where faster/smaller) is better.
>
> In that case I would go strictly with a socket-level approach.
>
> Just my 2 small coins,
> Elaine
>
>
>
> Vy Ho wrote:
> >
> > I look at a message send back from the server to a client in Axis, and
> > was very surprised that most of the text is not the data that I need to
> > send, but declaration and long namspaces. Is it possible to reduce
> > these down and still comply with Soap?
> >
> > Thanks,
> >
> > vh
> >
>
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> | Computers are useless. They can only give you answers.
> | -- Pablo Picasso --
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>
>
Re: Signal to noise ratio in Soap is very high
Posted by Elaine Nance <el...@commerce.state.ak.us>.
SOAP is not designed to create compact dataflows. It is designed
for interoperability and, because it is XML, extensibility.
IMHO, I would *not* use SOAP for purely internal processes if
speed or stream size matters (where faster/smaller) is better.
In that case I would go strictly with a socket-level approach.
Just my 2 small coins,
Elaine
Vy Ho wrote:
>
> I look at a message send back from the server to a client in Axis, and
> was very surprised that most of the text is not the data that I need to
> send, but declaration and long namspaces. Is it possible to reduce
> these down and still comply with Soap?
>
> Thanks,
>
> vh
>
<~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| Computers are useless. They can only give you answers.
| -- Pablo Picasso --
<~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
RE: Signal to noise ratio in Soap is very high
Posted by Christophe Roudet <cr...@activia.net>.
For the client side:
http://www.osmoticweb.com/axis-soap-compression.htm
also described in the previous link
http://www.osmoticweb.com/soap-compression-howto.htm
Christophe
> -----Original Message-----
> From: Vy Ho [mailto:st946tbf@drexel.edu]
> Sent: Tuesday, November 16, 2004 3:29 PM
> To: axis-user@ws.apache.org
> Subject: Re: Signal to noise ratio in Soap is very high
>
>
> Thank you for the link. Apache Tomcat already support server side gzip
> compression. I only need to do this on the client side. There's also
> free source code for doing the servlet that does the compression (very
> straight forward, because java already supply the gzip compression
> stream library). What left is the client side, which I could implement
> once knowing how extend this http handler thing..
>
> Christophe Roudet wrote:
>
> >Check http://www.osmoticweb.com/soap-compression-howto.htm
> >
> >Christophe
> >
> >
Re: Signal to noise ratio in Soap is very high
Posted by Vy Ho <st...@drexel.edu>.
Thank you for the link. Apache Tomcat already support server side gzip
compression. I only need to do this on the client side. There's also
free source code for doing the servlet that does the compression (very
straight forward, because java already supply the gzip compression
stream library). What left is the client side, which I could implement
once knowing how extend this http handler thing..
Christophe Roudet wrote:
>Check http://www.osmoticweb.com/soap-compression-howto.htm
>
>Christophe
>
>
RE: Signal to noise ratio in Soap is very high
Posted by Christophe Roudet <cr...@activia.net>.
Check http://www.osmoticweb.com/soap-compression-howto.htm
Christophe
> -----Original Message-----
> From: Vy Ho [mailto:st946tbf@drexel.edu]
> Sent: Tuesday, November 16, 2004 12:11 PM
> To: axis-user@ws.apache.org
> Subject: Re: Signal to noise ratio in Soap is very high
>
>
> Searching around, I found this article:
>
> http://www-106.ibm.com/developerworks/webservices/library/ws-
> sqzsoap.html#code2
>
> This basically add gzip to both end of the connection so that things
> gets smaller in the stream.
> The concept seems to be good. Is something Axis already implemented, so
> that I can just set a flag to say: use gzip on both the client and
> server? The idea of modifying Axis code doesn't sound well to me
> because with every new release of Axis, I have to again apply the patch.
Re: Signal to noise ratio in Soap is very high
Posted by Vy Ho <st...@drexel.edu>.
Searching around, I found this article:
http://www-106.ibm.com/developerworks/webservices/library/ws-sqzsoap.html#code2
This basically add gzip to both end of the connection so that things
gets smaller in the stream.
The concept seems to be good. Is something Axis already implemented, so
that I can just set a flag to say: use gzip on both the client and
server? The idea of modifying Axis code doesn't sound well to me
because with every new release of Axis, I have to again apply the patch.
Re: Signal to noise ratio in Soap is very high
Posted by Vy Ho <st...@drexel.edu>.
What I meant is the signal to noise ratio is very low.