If the entire MetaModel is preloaded, the time it takes to process the request is about a minute on a proper development machine. It only takes a few seconds to process the XML document and another 50+ seconds to process the XML document into the in-memory representation of the MetaModel.
This inefficiency persists at lower level but it is not as pronounced