In recent years, developing Modeling and Simulation (M&S) applications has become more and more complex. New technologies, like Web Services (WS) and Cloud computing, have been recently used in Modeling and Simulation (M&S). However, developing M&S applications using these technologies is still a complicated process. The reasons for this include: 1) it is hard to develop web services for varied M&S resources; 2) it is complicated to deploy M&S resources in the Cloud; 3) it is hard to integrate with varied M&S services; 4) it is complex to identify and select resources and services based on their meaning. In this research, we aim to simplify the development and integration of M&S applications using web technologies by solving the issues mentioned above. To do so, we propose the Mashup Architecture with Modeling and Simulation as a Service (termed MAMSaaS). MAMSaaS is a layered and lightweight M&S application development approach. It has five layers, which are Cloud, Box, Wiring, Mashup, and Tag Ontology Layers. It has a simplified life cycle to develop, deploy, identify, select, integrate and execute varied M&S resources as services in the Cloud. In the Cloud Layer, we developed CloudRISE middleware to expose RESTful Modeling and Simulation as a Service (MSaaS) for varied M&S resources; in addition, we propose new methods using Cloud computing and Experimental Framework concept to simplify the deployment of experiment environment. In the Box, Wiring and Mashup Layers, we present a new method based on mashup technologies to simplify the integration, execution and visualization of M&S applications. In the Tag Ontology Layer, we propose a new semantic selection approach using tag-mining and ontology-learning technologies, to identify and select M&S resources based on their meanings.