Parallel simulation and distributed simulation sometimes appear to be two different worlds. Where parallel simulation aims at increasing the speed of a single model by distributing it over more processors, distributed simulation looks at ways to link entire models or federates that run on different computers. In the case of distributed simulation, the models themselves are hard to distribute, and often they each run on one processor as a monolithic model. This paper advocates to build the more traditional simulation models in such a way, that they can be easily distributed. As simulationists, we can learn from component-based theory from the software engineering field to prepare our models for distribution, and parts of our models for reuse. The results of several projects show that componentizing simulation models can have many advantages. The results also show that it is not easy to create models in a componentized way, and that current methods for simulation model building should be adapted.