The conventional view of the congestion control problem in data networks is based on the principle that a flow’s performance is uniquely determined by the state of its bottleneck link, regardless of the topological properties of the network. However, recent work [1, 2, 3] has shown that the behavior of congestion-controlled networks is better explained by models that account for the web interactions among bottleneck links. These interactions are captured by a latent bottleneck structure, a model describing the complex ripple effects that changes in one part of the network exert on the other parts. In this presentation, we present GradientGraph, a new network optimization platform that leverages bottleneck structure analysis to detect, characterize, predict and resolve network bottlenecks in high-speed data networks. GradientGraph from Reservoir Labs provides a new paradigm to qualitatively and quantitatively understand network bottlenecks and flows, enabling applications including real-time traffic engineering, optimized routing and flow control, predictive modeling, and a framework for high-precision network design and capacity planning, among others. GradientGraph Analytics provide a cost-effective way for network operators to optimize their network for reliable and predictable service.
 Jordi Ros-Giralt, Atul Bohara, Sruthi Yellamraju, Harper Langston, Richard Lethin, Yuang Jiang, Leandros Tassiulas, Josie Li, Ying Lin, Yuanlong Tan, Malathi Veeraraghavan, “On the Bottleneck Structure of Congestion-Controlled Networks,” ACM SIGMETRICS, Boston, June 2020.
 Jordi Ros-Giralt, Noah Amsel, Sruthi Yellamraju, James Ezick, Richard Lethin, Yuang Jiang, Aosong Feng, Leandros Tassiulas, Zhenguo Wu, Min Yeh Teh, Keren Bergman, “Designing Data Center Networks Using Bottleneck Structures,” ACM SIGCOMM, August 2021. (Accepted for publication.)
 Noah Amsel, Jordi Ros-Giralt, Sruthi Yellamraju, Brendan von Hofe, Richard Lethin, “Computing Bottleneck Structures at Scale for High-Precision Network Performance Analysis,” International Workshop on Innovating the Network for Data Intensive Science (INDIS), Supercomputing, Denver, Nov 2020.
For more information or a copy of this paper, please contact us.