Internet congestion control mechanisms are critical for managing network traffic and ensuring efficient resource utilization. This article reviews the current Transmission Control Protocol (TCP) congestion control protocols and highlights recent analytical advances that have enabled a deeper understanding of these mechanisms. The paper presents an optimization-based framework that interprets various flow control mechanisms, particularly the utility being optimized by the protocol's equilibrium structure. It also examines the dynamics of TCP and employs linear models to analyze stability limitations in predominant TCP versions, despite built-in compensations for delay. A new protocol is proposed that overcomes these limitations and provides stability in a scalable manner for arbitrary networks, link capacities, and delays. TCP, which uses "window" flow control, adjusts the rate of data transmission based on network feedback. The predominant TCP implementations, Tahoe and Reno, use a combination of linearly increasing and exponentially decreasing window sizes to manage congestion. Reno introduces improvements such as fast recovery to more efficiently handle packet loss. TCP Vegas improves upon Reno by using queueing delay as a congestion measure, leading to more stable and efficient performance. The paper discusses various congestion control mechanisms, including FIFO, DropTail, and RED, which manage network congestion through different strategies. Analytical models are developed to understand the behavior of these mechanisms, focusing on equilibrium properties and the optimization of utility functions. These models show that congestion control can be interpreted as an economic problem, where sources aim to maximize their utility while considering the costs of congestion. The paper also addresses the stability of TCP protocols, particularly Reno/RED, and the challenges posed by network delays and varying link capacities. It proposes a new protocol that can be implemented in a decentralized manner, providing linear stability for arbitrary delays, capacities, and routes. The analysis shows that while TCP Reno has performed well, it has limitations in handling certain types of traffic, particularly those with bursty characteristics. The paper argues for a more analytical approach to congestion control, leveraging recent advances in modeling and control theory to develop more efficient and stable protocols.Internet congestion control mechanisms are critical for managing network traffic and ensuring efficient resource utilization. This article reviews the current Transmission Control Protocol (TCP) congestion control protocols and highlights recent analytical advances that have enabled a deeper understanding of these mechanisms. The paper presents an optimization-based framework that interprets various flow control mechanisms, particularly the utility being optimized by the protocol's equilibrium structure. It also examines the dynamics of TCP and employs linear models to analyze stability limitations in predominant TCP versions, despite built-in compensations for delay. A new protocol is proposed that overcomes these limitations and provides stability in a scalable manner for arbitrary networks, link capacities, and delays. TCP, which uses "window" flow control, adjusts the rate of data transmission based on network feedback. The predominant TCP implementations, Tahoe and Reno, use a combination of linearly increasing and exponentially decreasing window sizes to manage congestion. Reno introduces improvements such as fast recovery to more efficiently handle packet loss. TCP Vegas improves upon Reno by using queueing delay as a congestion measure, leading to more stable and efficient performance. The paper discusses various congestion control mechanisms, including FIFO, DropTail, and RED, which manage network congestion through different strategies. Analytical models are developed to understand the behavior of these mechanisms, focusing on equilibrium properties and the optimization of utility functions. These models show that congestion control can be interpreted as an economic problem, where sources aim to maximize their utility while considering the costs of congestion. The paper also addresses the stability of TCP protocols, particularly Reno/RED, and the challenges posed by network delays and varying link capacities. It proposes a new protocol that can be implemented in a decentralized manner, providing linear stability for arbitrary delays, capacities, and routes. The analysis shows that while TCP Reno has performed well, it has limitations in handling certain types of traffic, particularly those with bursty characteristics. The paper argues for a more analytical approach to congestion control, leveraging recent advances in modeling and control theory to develop more efficient and stable protocols.