Alibaba’s Contributions and Vision at SIGCOMM 2018: Future Network Technologies

On August 21, 2018, SIGCOMM 2018, the premier academic conference on future network technologies, was held in Hungary and attracted attention from researchers and industry leaders across Asia, North America, and Europe. Alibaba participated for the first time, presenting its network technology roadmap and groundbreaking advances, which received high recognition from attendees.

SIGCOMM is the top academic venue in networking, known for its rigorous paper selection—only about 30 papers are accepted each year based on fundamental contributions, forward‑looking impact, and solid system implementations. Publishing at SIGCOMM is a coveted honor for network researchers.

This year, Alibaba was the only Chinese company with technical experts involved in paper reviewing and shaping the conference direction.

On the first day of the conference, Alibaba hosted an "Alibaba Night" event, sharing its future network vision and challenges with professors and PhD students from leading institutions such as MIT, Stanford, Carnegie Mellon, UC Berkeley, and Princeton. Notable participants included SIGCOMM Lifetime Achievement Award winner Prof. Jennifer Rexford, Tsinghua Professor Bijun Bi, CMU Prof. Justin Sherry, and Cornell Prof. Rachit Agarwal.

Alibaba highlighted three major directions for future network breakthroughs: programmable/custom hardware (e.g., SmartNICs, programmable ASICs), integrated software‑hardware design (e.g., RDMA, kernel bypass), and intelligent network operations (e.g., AIOps, intent‑based networking).

In these areas, Alibaba has already conducted systematic, forward‑looking research, with several results deployed at scale on its production network.

Alibaba’s chief network architect, Cai Dezhi, presented the "Hyper‑scale Edge Network" vision, describing how cloud computing will move closer to users, deploying massive edge networks to support IoT and other latency‑sensitive applications. The vision emphasizes three core challenges: efficient automated operations, intelligent traffic scheduling, and multi‑layer security.

During the Self‑Driving Networks Workshop chaired by Prof. Jennifer Rexford, Alibaba presented the paper "Automatic Life Cycle Management of Network Configurations," introducing a "network hierarchical model" that enables fully automated configuration generation, verification, and changes across Alibaba’s global network, elevating network intelligence to unprecedented levels.

Alibaba also showcased its leadership in high‑performance networking. It is one of the few companies that have deployed RDMA at large scale in production. At SIGCOMM, senior expert Dr. Liu Hongqiang organized the Kernel Bypassing Networks Workshop, featuring a keynote by Mellanox co‑founder Dr. Eitan Zahavi, discussing next‑generation RDMA hardware and flow‑control algorithms.

In network visualization, Alibaba pioneered large‑scale deployments, collaborating with top hardware vendors to design new architectures and interfaces. The company demonstrated intelligent network design, sub‑second fault recovery, and next‑generation data‑center traffic scheduling, significantly improving bandwidth utilization and stability.

Paper materials shared at the conference included:

"Automatic Life Cycle Management of Network Configurations" "Elastic Sketch: Adaptive and Fast Network‑Wide Measurements" "Network Transparentization in Alibaba"

For more details, visit the SIGCOMM 2018 website: https://conferences.sigcomm.org/sigcomm/2018

Key Alibaba contributors mentioned were Dr. Zhang Ming (researcher, SIGCOMM paper reviewer and Network Verification Session chair) and Dr. Liu Hongqiang (senior technical expert, KBNets Workshop chair).