Almost all the ideas and things which have a suffix or prefix “framework” are basically composite models by design i.e. multi-variable models and not the single variable models. This include FEAF, DODAF, RM-ODP, Gartner EA, Forrester EA, Microsoft EA, RUP / 4+1, EUP, Balanced Score Card, ComVantage etc. etc. Further, all these framework mandates certain number of “models” to be created without realizing that those models are insufficient, incomplete and inconsistent in terms of supporting six variables (what, how, where, when, who, why) and six perspectives (identification, definition, representation, specifications, implementation/configuration, instantiation).

As John Zachman defines- single variable models are Architecture models. There are six variables, i.e. data, function, network, role, time and rules. If a model consists of a single variable type, then it's Architecture model. However, if a model has more than one variable, say a model is having reference to data & function, or other model having a function, data and role references, etc. these are examples of composite models (implementation models). Interestingly, in most cases, models are multi-variable and hence an implementation model. As the number of variables is added to a single diagram, they tend to become less legible and these models more likely to change quickly and become outdated rendering them useless owing to their shorter life span.