Due to the lack of a general theoretical foundation, today's distributed traffic control mechanisms developed at the networking layer, transport layer, and overlay are largely disintegrated. As a result, traffic control protocols developed at different layers may achieve conflicting design objectives and interact with one another in an unpredictable fashion. In this paper, we propose a novel strategy to tackle this issue. First, we propose a theoretical foundation for distributed traffic control. On the basis of this foundation, we then propose an integrated, multilayer, multi-domain traffic control structure. This structure makes it possible to develop traffic control protocols at different layers, possessing the following nice features: (1) they achieve non-conflicting design objectives; (2) they enable rich service quality features, including Quality-of-Service (QoS), Traffic Engineering (TE), and Fast Failure Recovery (FFR); (3) they lead to highly scalable, globally stable and optimal control; (4) they can deal with network diversities and tussles among administrative domains; (5) they allow effective control of dynamically generated overlay networks. The proposed strategy only makes two assumptions about the Internet architecture, i.e., the ability to support multiple domains and multi-path forwarding. As a result, the proposed strategy can be applied to the existing or any future Internet architectures for which these two assumptions hold.