Skip to content.

Manageability

Sections
Personal tools
You are here: Home » blog » archive » Slaying the XML Behemoth

Slaying the XML Behemoth

20030603070025

Dennis Sosnoski has been making a living out of building scalable solutions for XML.   He's developed XML Bench a set of benchmarks for different XML models in Java.  He's developed XML Stream a highly effecient serialization of XML.

Now, he's put together an open source package to efficiently handle XML binding (i.e. translation of XML to and from Java).  See, one of the biggest, if not the biggest problem, of XML is that its just to easy to create extremely humongous documents.  In my last project, I've seen documents reaching almost 70 Megabytes in size.  Parsing and just holding on to a document of this kind can bring a JVM to its knees and in most cases kill that JVM.   So, if your in a project that uses XML in a heavy way and you haven't adressed the scalability the problem, then your setting yourself up for a big shock.

Sosnoksi has put together JIBX.  It's an XML binding framework that addresses a insightful question "Why should XML binding implementations consume more memory than XML parsers?".  Afterall, a translation to a Java object ignores most of the text found in an XML document, it should then require a much lower working set.  (note: Although it's not a branded standard like JAXB, the success of your project may ultimately rely on whether your server stays up or not.  In this case conforming to standards is a moot point.)

Amidst all the hype of using XML, there's a serious need to make this scalable.  That's what Dennis Sosnoski is doing and I'm hoping that more in the industry would address this problem instead of sweeping it under the rug as a disinteresting implementation concern.

Created by admin
Last modified 2003-07-30 04:15 PM
 

Powered by Plone

This site conforms to the following standards: