W3C home > Mailing lists > Public > xml-dist-app@w3.org > July 2001

Re: benchmarking SOAP?

From: Aleksander Slominski <aslom@watson.ibm.com>
Date: Wed, 25 Jul 2001 16:39:40 -0400
Message-ID: <02d801c11549$ed7f3fa0$ac130209@watson.ibm.com>
To: "Syed Ahmad" <sahmad@netspeak.com>, "'Aleksander Slominski'" <aslom@cs.indiana.edu>
Cc: "XML Distributed Applications List" <xml-dist-app@w3.org>
> Thanks for the info. This is very helpful. One of the issues I have seen
is
> large XML data (imagine XML based data about provisioning the phone record
> of say 20,000 subscribers - pushed at 3:00 a.m. Sunday morning) being
> compressed before it is transported. The SOAP method headers can stay
> uncompressed. In this case - there is also performance improvements.
> Ofcourse there is slight delay in packing the XML data and then unpacking
> them at the destination.
>
> Have u done some testing with packed XML data as well ?

hi,

we were doing tests on LAN as we were interested in XML parsing speed (and
in this case compression actually slowed down round-trip time as bottleneck
was in XML parsing which is CPU intensive).

it would be interesting to run some tests to find "sweet point" for using
compression over WAN - both compression and XML parsing is CPU intensive so
compressing message will introduce overhead especially for very small XML
messages that even uncompressed fits into one network packet. however
obviously compression pays off when transmitting bigger messages (and
receiving them over modem...)

thanks,

alek
Received on Wednesday, 25 July 2001 16:40:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:03 GMT