This paper introduces the "allocated-capacity" framework, which aims to provide different levels of best-effort service during network congestion. The framework extends Internet protocols and algorithms to allocate bandwidth to users in a controlled and predictable manner. It supports two complementary control methods: sender-based and receiver-based. The framework can be used for charging and more efficiently utilizing network resources in heterogeneous and commercial Internet environments. The paper focuses on algorithms for essential components, including a differential dropping algorithm for routers and a tagging algorithm for profile meters at the network edge for bulk-data transfers. Simulation results demonstrate the effectiveness of these algorithms in controlling TCP traffic to achieve targeted sending rates. The framework also provides a simple way to identify nonresponsive users at aggregation points.This paper introduces the "allocated-capacity" framework, which aims to provide different levels of best-effort service during network congestion. The framework extends Internet protocols and algorithms to allocate bandwidth to users in a controlled and predictable manner. It supports two complementary control methods: sender-based and receiver-based. The framework can be used for charging and more efficiently utilizing network resources in heterogeneous and commercial Internet environments. The paper focuses on algorithms for essential components, including a differential dropping algorithm for routers and a tagging algorithm for profile meters at the network edge for bulk-data transfers. Simulation results demonstrate the effectiveness of these algorithms in controlling TCP traffic to achieve targeted sending rates. The framework also provides a simple way to identify nonresponsive users at aggregation points.