During 1988-2008, research in networks coalesced around cross-layer resource optimization and QoS-driven design. Generalized processor sharing with leaky-bucket admission control established a fair, service-rate sharing framework suitable for multi-class traffic, while dynamic, receiver-initiated resource reservation with soft-state management enabled flexible QoS in IP networks. In wireless and wired settings, stability analyses and throughput-optimal scheduling for multi-hop systems advanced cross-layer design, and the equivalent capacity concept provided a practical provisioning metric under traffic fluctuations. Overall, the period emphasized adaptive, distributed, and scalable resource management across layers, bridging theoretical throughput results with practical QoS implementations. Historical Significance: The era cemented cross-layer resource optimization as a central paradigm, shaping both theoretical and applied directions. The introduction of throughput-optimal scheduling, the emergence of receiver-initiated reservation principles, and the soft-state approach to QoS management became foundational components of later network architectures. Foundational ideas like generalized service-rate sharing, stability conditions for complex networks, and capacity-provisioning metrics influenced subsequent developments in IP QoS, cross-layer wireless design, and large-scale capacity planning, illustrating a shift from static provisioning to dynamic, performance-driven resource management.