Re: Any best practise to make CXF better support large amount data transmit

[Jessie914 <sh...@yahoo.com.cn>]

hi, Glen

Many thanks for the reply, I am now able to invoke async calls.

Regarding to the MTOM,  if MTOM is used, the server side must enable the
MTOM as well, correct?  

if there is nothing I can change on the server side,  the only transport way
is to use the default way?

thanks again.


Glen Mazza wrote:
> 
> No, the web service provider shouldn't need to do anything special in
> order for you to make an asynchronous web service call.  Async[1] will
> help you if the call will take a long time to process (and the SOAP client
> can do something else while it's waiting), not necessarily in cases where
> you're just returning lots of data--the two cases are similar but not
> exactly the same.
> 
> If the data in the SOAP response can be more efficiently stored as binary
> instead of XML, MTOM[2] might be worth considering.
> 
> Glen
> 
> [1]
> http://www.jroller.com/gmazza/entry/creating_service_side_asynchronous_web
> (first paragraph for samples)
> [2] http://www.jroller.com/gmazza/entry/using_mtom_and_apache_fop
> 
> 
> Jessie914 wrote:
>> 
>> Hello,
>> 
>> I am a newbie to CXF. We have a web service server implemented by CXF
>> V2.1 (which cannot be changed), and now we are at the begining to develop
>> the web service client. The web service call will return large amount
>> data.
>> 
>> Will Aysnc calls help in this case?  I am not sure if Aysnc mechanism
>> already implemented on the server side. (it is a prereq for client to
>> implement aysnc calls, isn't it?)   
>> 
>> Any best practise to improve the web service performance?
>> 
>> Many thanks.
>> regards.
>>    
>> 
> 
> 

-- 
View this message in context: http://old.nabble.com/Any-best-practise-to-make-CXF-better-support-large-amount-data-transmit-tp26569310p26588097.html
Sent from the cxf-user mailing list archive at Nabble.com.

Re: Any best practise to make CXF better support large amount data transmit

[Daniel Kulp <dk...@apache.org>]

On Tue December 1 2009 4:16:28 am Jessie914 wrote:
> hi, Glen
> 
> Many thanks for the reply, I am now able to invoke async calls.
> 
> Regarding to the MTOM,  if MTOM is used, the server side must enable the
> MTOM as well, correct?

If you want the server to RETURN MTOM, yes.   The server (and client) will 
always accept MTOM even if mtom is turned "off".   It just won't produce MTOM.


> if there is nothing I can change on the server side,  the only transport
>  way is to use the default way?

Pretty much.   Being stuck on 2.1 also limits things a bit.     With 2.2.5 and 
such, we support streaming attachments much better.    We also can have gzip 
encoding enabled with the mtom stuff, thus reducing bandwidth even more.

Dan




> 
> thanks again.
> 
> Glen Mazza wrote:
> > No, the web service provider shouldn't need to do anything special in
> > order for you to make an asynchronous web service call.  Async[1] will
> > help you if the call will take a long time to process (and the SOAP
> > client can do something else while it's waiting), not necessarily in
> > cases where you're just returning lots of data--the two cases are similar
> > but not exactly the same.
> >
> > If the data in the SOAP response can be more efficiently stored as binary
> > instead of XML, MTOM[2] might be worth considering.
> >
> > Glen
> >
> > [1]
> > http://www.jroller.com/gmazza/entry/creating_service_side_asynchronous_we
> >b (first paragraph for samples)
> > [2] http://www.jroller.com/gmazza/entry/using_mtom_and_apache_fop
> >
> > Jessie914 wrote:
> >> Hello,
> >>
> >> I am a newbie to CXF. We have a web service server implemented by CXF
> >> V2.1 (which cannot be changed), and now we are at the begining to
> >> develop the web service client. The web service call will return large
> >> amount data.
> >>
> >> Will Aysnc calls help in this case?  I am not sure if Aysnc mechanism
> >> already implemented on the server side. (it is a prereq for client to
> >> implement aysnc calls, isn't it?)
> >>
> >> Any best practise to improve the web service performance?
> >>
> >> Many thanks.
> >> regards.
> 

-- 
Daniel Kulp
dkulp@apache.org
http://www.dankulp.com/blog