Quantitative end-to-end throughput differentiation for bulk TCP applications in the Internet

The vast majority of popular Internet applications are controlled by TCP. The provision of end-to-end quantitative QoS guarantees to individual TCP flows is thus an exceedingly interesting problem, facilitating new business models for the Internet and therefore encouraging investments on the part of Internet Service Providers (ISPs). However, although this field has been extensively researched in previous decades, the failure of the resulting proposals to deploy Internet-wide in any meaningful way is especially noteworthy. This failure showcases the importance that scalability and implementation complexity issues can have in the success of Internet-wide QoS technologies.In this paper, we present a scalable, robust and exceedingly simple method able to provide end-to-end quantitative guarantees for the throughput of individual bulk TCP flows in the Internet. Our method does not mandate changes to existing Internet infrastructure nor complex agreements among ISPs, making its implementation straightforward in the current Internet.