The complexity of services provided through theWeb is continuously increasing and issues introduced by both heterogeneous client devices and Web content personalization are becoming a major challenge for the Web. The content adaptation operations may be computationally expensive, requiring high efficiency and scalability in the Web architectures. Tailoring Web and multimedia resources to meet the user and client requirements opens two main novel issues in the research area of content adaptation and delivery. First, the characteristics of the content adaptation services have a significant impact on the possible mapping of the services over the nodes of the content adaptation infrastructure. Second, the relationship of the content adaptation system with the content provider has a fundamental impact on the architectural choices in the design of content adaptation systems.

In this paper, we propose four novel distributed architectures for the efficient delivery of content adaptation services where the nodes are organized in two levels. We compare the performance of the four architectures to evaluate which architecture is the best and under which conditions the performance gain is more evident.