Yes, that's exactly why I created the LazyTreeModel. The DefaultTreeModel works best when the entire model is loaded into memory and the model is stored between requests (in the session or as a singleton). The first time a node is expanded, it must be found by traversing from the root which has complexity O(N). Subsequent expansions of the same node have complexity O(1) since a hash lookup in a cache is done.
In contrast, the LazyTreeModel has complexity of O(1) without needing to be stored between requests. For this reason, I feel it is more scalable since you never store the entire tree in memory.
