As someone who is also working in the wild with WS and SOAP I'd say that the
problems I've seen around performance have been the age old ones of

1) Fine grained interactions
2) Chatty interactions
3) "Ship it all, they'll work out what they need"
4) Rubbish performance behind the interface being blamed on the interface

Now clearly SOAP and XML marshalling aren't free, but when we did some
RMI/SOAP tests back in 2003/4 we found that RMI/IIOP v SOAP was a pretty
even fight, now I'll give Dan and Greg the chance to jump on that as being
the wrong comparison but it was fair for that project.  My first production
experience of SOAP was in 2001 over a 1 MB SSL over the internet link
running off 2 way Intel boxes, it worked for that fine and I'm confused as
to why performance has got worse since then.

So is the issue with performance actually performance of SOAP, or is it that
people aren't good at building distributed applications.

NB to the REST people out there, don't snigger around this as the chatty
nature of REST is liable to create equally large problems against the weight
of SOAP.

Steve


On 17/01/07, Gervas Douglas <[EMAIL PROTECTED]> wrote:

  <<For years those building SOAs have said that "SOAP is too slow" and
Web Services are just the icing on the SOAP cake. However, as somebody
who's out there in the SOA project world right now I think it's fair
to say that many SOAs are indeed slow, and that performance is always
an issue when dealing with Web services.

However, SOAP could be speeding up. Enter WS-MTOM, or the development
of the SOAP Message Transmission Optimization Mechanism specification.
MTOM offers composability of base64 with the transport efficiency of
SOAP with attachments. However, WS-MTOM wasn't tied into the rest of
the Web Services architecture: there was no standard way for services
to advertise that they were "MTOM ready," until now.

IBM and Microsoft have recently submitted WS-MTOMPolicy to W3C. This
has now been acknowledged by W3C, which clears the way for a
standardization effort around this issue.

From the spec:

"This specification describes a domain-specific policy assertion
that indicates endpoint support of the optimized MIME multipart/
related serialization of SOAP messages defined in section 3 of the
SOAP Message Transmission Optimization Mechanism specification. This
policy assertion can be specified within a policy alternative as
defined in WS-Policy Framework and attached to a WSDL description as
defined in WS-PolicyAttachment."

So, take WS-Policy and MTOM, and it's soon going to be possible for
Web Services across enterprise boundaries to advertise their MTOM
capabilities. Thus, SOAP will be faster. Thus, Web services will be
faster. Thus, our SOAs will be faster. That's a good thing.>>

You can read this at:

<
http://weblog.infoworld.com/realworldsoa/archives/2007/01/speed_up_your_s.html?source=NLC-SOA2007-01-18
>

Gervas

Reply via email to