Another approach is to create a data structure that dynamically changes based on how the node is populated. E.g., a CacheDataStructure may be a very light weight wrapper, which could create and hold a reference to a CHM only if the number of attributes crosses a threshold, and use simple object references for smaller numbers of attributes. I've seen a lot of use cases where a node holds very few (<3) attribs, often only one. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4077021#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...