We present a novel bandwidth broker architecture for scalable support of guaranteed services that decouples the QoS control plane from the packet forwarding plane. More specifically, under this architecture, core routers do not maintain any QoS reservation states, whether per-flow or aggregate. Instead, QoS reservation states are stored at and managed by bandwidth broker(s). There are several advantages of such a bandwidth broker architecture. Among others, it relieves core routers of QoS control functions such as admission control and QoS state management, and thus enables a network service provider to introduce new (guaranteed) services without necessarily requiring software/hardware upgrades at core routers. Furthermore, it allows us to design efficient admission control algorithms without incurring any overhead at core routers. The proposed bandwidth broker architecture is designed based on a core stateless virtual time reference system developed in . In this paper we focus on the design of efficient admission control algorithms under the proposed bandwidth broker architecture. We consider both per-flow guaranteed delay services and class-based guaranteed delay services with flow aggregation. We demonstrate how admission control can be done on an entire path basis, instead of on a `hop-by-hop' basis. Such an approach may significantly reduce the complexity of the admission control algorithms. We also study the impact of dynamic flow aggregation on the design of class-based admission control algorithms. Based on the proposed bandwidth broker architecture, we devise effective mechanisms to circumvent the problem caused by dynamic flow aggregation.