Maestro has the concept of 'deferred instantiation' or 'lazy loading'
What are the technical details, and what is the comparative performance given the maestro architecture and that of the server component in TM 5.1 vs that of Composer (large form, 20 pages for example)
With Maestro the entire application definition is downloaded with the application. However the pages are only loaded (created) on demand this provides substantial performance improvements with large applications.
For the end user this means applications will be available to use much sooner, the difference can be in the order of many seconds.