TY - GEN
T1 - Dragonfly
T2 - 22nd International Federation for Information Processing Conference on Networking, IFIP Networking 2023
AU - Carmel, Dean
AU - Keslassy, Isaac
N1 - Funding Information: ACKNOWLEDGMENT The authors would like to thank Roy Mitrany for his crucial help with the implementation. This work was partly supported by the Louis and Miriam Benjamin Chair in Computer-Communication Networks, the Israel Science Foundation (grant No. 1119/19), Toga Networks, and the Hasso Plattner Institute Research School. Publisher Copyright: © 2023 IFIP.
PY - 2023
Y1 - 2023
N2 - We introduce the Dragonfly system, which is designed to classify on the fly the congestion control algorithm of any flow that crosses a given router, starting at any time, and quickly reach a reasonable accuracy. To do so, we discuss the unique challenges of real-time congestion control classification. We explain how the number of bytes of the flow within the shared router queue contains an intrinsic memory that significantly helps real-time classification. However, we show that this number of bytes is not straightforward to compute in real time, and introduce ways to do so. We further design an eBPF-based scalable traffic-collection system that helps dynamically filter specific flows at high rates. Finally, we evaluate our Dragonfly system using a variety of platforms, and show that it clearly outperforms state-of-the-art algorithms.
AB - We introduce the Dragonfly system, which is designed to classify on the fly the congestion control algorithm of any flow that crosses a given router, starting at any time, and quickly reach a reasonable accuracy. To do so, we discuss the unique challenges of real-time congestion control classification. We explain how the number of bytes of the flow within the shared router queue contains an intrinsic memory that significantly helps real-time classification. However, we show that this number of bytes is not straightforward to compute in real time, and introduce ways to do so. We further design an eBPF-based scalable traffic-collection system that helps dynamically filter specific flows at high rates. Finally, we evaluate our Dragonfly system using a variety of platforms, and show that it clearly outperforms state-of-the-art algorithms.
UR - https://www.scopus.com/pages/publications/85167866007
U2 - 10.23919/IFIPNetworking57963.2023.10186432
DO - 10.23919/IFIPNetworking57963.2023.10186432
M3 - Conference contribution
T3 - 2023 IFIP Networking Conference, IFIP Networking 2023
BT - 2023 IFIP Networking Conference, IFIP Networking 2023
Y2 - 12 June 2023 through 15 June 2023
ER -