This paper analyzes the end-to-end packet delay and loss behavior in the Internet using measurements of small UDP probe packets sent at regular intervals. By varying the interval between packets, the study examines the structure of Internet load over different time scales, ranging from a few milliseconds to a few minutes. The observations align with previous findings from simulations and experiments, confirming the hypothesis of a mix of bulk traffic with larger packet sizes and interactive traffic with smaller packet sizes. The study also reveals interesting behaviors, such as the random nature of packet losses unless the probe traffic uses a significant fraction of the available bandwidth. The results have implications for the design of control mechanisms in the Internet, particularly for audio and video applications, where understanding packet delay and loss characteristics is crucial. The paper concludes by discussing the potential of using a simple single-server queueing model to analyze these behaviors and the ongoing research efforts to refine this model.This paper analyzes the end-to-end packet delay and loss behavior in the Internet using measurements of small UDP probe packets sent at regular intervals. By varying the interval between packets, the study examines the structure of Internet load over different time scales, ranging from a few milliseconds to a few minutes. The observations align with previous findings from simulations and experiments, confirming the hypothesis of a mix of bulk traffic with larger packet sizes and interactive traffic with smaller packet sizes. The study also reveals interesting behaviors, such as the random nature of packet losses unless the probe traffic uses a significant fraction of the available bandwidth. The results have implications for the design of control mechanisms in the Internet, particularly for audio and video applications, where understanding packet delay and loss characteristics is crucial. The paper concludes by discussing the potential of using a simple single-server queueing model to analyze these behaviors and the ongoing research efforts to refine this model.