Martin, About discrepancies found in timings between your and Eric's test, could possibly attributed to the type of payload. When we attempted sending an html payload over SOAP body, SOAP started parsing the html as well instead of ignoring the tags and treated it as character data. Payload size was about 70K.
A higher version of xerces parsers (1.4+) brought the response time down from 80 seconds to 20. -Sharmin. >Mailing-List: contact [EMAIL PROTECTED]; run by ezmlm >list-help: <mailto:[EMAIL PROTECTED]> >list-unsubscribe: <mailto:[EMAIL PROTECTED]> >list-post: <mailto:[EMAIL PROTECTED]> >Delivered-To: mailing list [EMAIL PROTECTED] >From: "Martin Jericho" <[EMAIL PROTECTED]> >To: <[EMAIL PROTECTED]> >Subject: Re: sending large files >Date: Mon, 30 Sep 2002 08:22:09 +1000 >MIME-Version: 1.0 >Content-Transfer-Encoding: 7bit >X-Priority: 3 >X-MSMail-Priority: Normal >X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4910.0300 >X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N > >Hi Tom > >My comment about no-one seeming to be interested wasn't meant to be taking a >swipe at anyone, I was just surprised that I had received no response at >all. I'm not expecting an immediate investigation, but normally if you get >no response, it means your question is too dumb or no-one is interested. I >really just wanted to find out whether it's a known problem, and if not, to >make it known. > >My standard answer to "Sooner, if you help out", is that I simply don't have >the time to get that involved in another project when I already have such >tight deadlines. I have to break up my message into chunks anyway because >of the lack of support for streaming, so it is not really a problem for me >personally. > >As it turns out, the Eric's problem may be something else again. He was >talking about times of over 15 minutes with an 800K message, but my test >showed a complete turnaround time of 20.7 seconds with a single message of >about 800K. > >Martin > >----- Original Message ----- >From: "Tom Jordahl" <[EMAIL PROTECTED]> >To: <[EMAIL PROTECTED]> >Sent: Saturday, September 28, 2002 4:04 AM >Subject: RE: sending large files > > >> >> Martin, >> >> We are all very interested in performance for Axis. >> >> What we would be MORE interested in, is if you were to pinpoint WHERE and >WHY we have the problems that led to your findings. Bonus points for >providing a PATCH to fix the problems. >> >> The standard answer to the question: "When will X be done/fixed?" >> "Sooner, if you help out" >> >> -- >> Tom Jordahl >> Macromedia Server Development >> >> >> >> -----Original Message----- >> From: Martin Jericho [mailto:[EMAIL PROTECTED]] >> Sent: Thursday, September 26, 2002 6:21 PM >> To: [EMAIL PROTECTED] >> Subject: Re: sending large files >> >> >> >> ----- Original Message ----- >> From: "James Black" <[EMAIL PROTECTED]> >> To: <[EMAIL PROTECTED]> >> Sent: Friday, September 27, 2002 7:28 AM >> Subject: Re: sending large files >> >> >> > "Jung, Eric (Contractor)" wrote: >> > >> > > Have you turned off debugging output? With a stdout appender, things >are >> > > extremely slow for me with message sizes no where approaching 800K. >> > >> > Turned it off where? >> >> org.apache.log4j.Category.getDefaultHierarchy().disableDebug(); >> >> > I am installing Tomcat 4.1.12 at the moment,but would like to know how >> to >> > turn off any debugging also. >> >> Regarding the performance problem, you should indeed half the time >required >> by turning off debugging, but the real problem is worse than that. I >posted >> a message to this mailing list a couple of days ago (Subject: Performance >> problems with RPC messages over 20k), where I did some benchmarks based on >> message size. It looks like the time required to process the message >> increases at a geometric rate relative to the message size. >> >> As yet no-one else seems to be interested in this problem, so the best >thing >> I can suggest is to break your string up into 20K chunks. >> >> Please let me know whether you find out anything more about it. >> >> Martin Jericho >> >
