Distancer: A Host-Based Distributed Adaptive Load Balancer for Datacenter Traffic

  • Songyun Wang
  • Xin LiEmail author
  • Zhuzhong Qian
  • Jiabin Yuan
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 11335)


Contemporary datacenter networks are typically organized with multi-rooted tree topologies. To fully utilize the multiple end-to-end paths, effective mechanisms are required to balance traffic across them. However, existing load balancers for datacenters either operate at a coarse granularity, or support little for network failures, or necessitate customized hardware. We propose Distancer, a host-based distributed adaptive load balancer for datacenter traffic, which requires no coordination and modification of switches. Based on a deep investigation of TCP feedback mechanism, we firstly design Congestion Detector (C-Detector), which exploits ACKs to effectively handle network hot-spots and path anomalies in real time; Then we develop Load-Balancer (L-Balancer) to select best paths for both data packets and ACKs. According to our extensive evaluations, Distancer can achieve up to 40% and 20% better average flow completion times (AFCTs) than ECMP and CONGA respectively. Under the presence of path failures, Distancer improves the AFCT up to 400% and 30% over ECMP and CONGA.


Data center networking Flow scheduling Load balance 


  1. 1.
  2. 2.
    The NS-2 network simulator.
  3. 3.
    Al-Fares, M., Loukissas, A., et al.: A scalable, commodity data center network architecture. In: ACM SIGCOMM CCR, vol. 38, pp. 63–74 (2008)CrossRefGoogle Scholar
  4. 4.
    Al-Fares, M., Radhakrishnan, S., et al.: Hedera: dynamic flow scheduling for data center networks. In: Proceedings of NSDI, vol. 10, p. 19 (2010)Google Scholar
  5. 5.
    Alizadeh, M., Edsall, T., et al.: CONGA: distributed congestion-aware load balancing for datacenters. In: Proceedings of ACM SIGCOMM, pp. 503–514 (2014)CrossRefGoogle Scholar
  6. 6.
    Benson, T., Akella, A., et al.: Network traffic characteristics of data centers in the wild. In: Proceedings of ACM IMC, pp. 267–280 (2010)Google Scholar
  7. 7.
    Benson, T., Anand, A., et al.: MicroTE: fine grained traffic engineering for data centers. In: Proceedings of ACM CoNEXT, p. 8 (2011)Google Scholar
  8. 8.
    Cao, J., Xia, R., et al.: Per-packet load-balanced, low-latency routing for clos-based data center networks. In: Proceedings of ACM CoNEXT, pp. 49–60 (2013)Google Scholar
  9. 9.
    Cao, Y., Xu, M., et al.: Explicit multipath congestion control for data center networks. In: Proceedings of ACM CoNEXT, pp. 73–84 (2013)Google Scholar
  10. 10.
    Cheung, C.M., Leung, K.C.: DFFR: a flow-based approach for distributed load balancing in data center networks. Comput. Commun. 116, 1–8 (2018)CrossRefGoogle Scholar
  11. 11.
    Dixit, A., Prakash, P., et al.: On the impact of packet spraying in data center networks. In: Proceedings of IEEE INFOCOM, pp. 2130–2138 (2013)Google Scholar
  12. 12.
    Gill, P., Jain, N., et al.: Understanding network failures in data centers: measurement, analysis, and implications. In: ACM SIGCOMM CCR, vol. 41, pp. 350–361 (2011)CrossRefGoogle Scholar
  13. 13.
    Greenberg, A., Hamilton, J.R., et al.: Vl2: a scalable and flexible data center network. In: ACM SIGCOMM CCR, vol. 39, pp. 51–62 (2009)CrossRefGoogle Scholar
  14. 14.
    Guo, C., Lu, G., et al.: SecondNet: a data center network virtualization architecture with bandwidth guarantees. In: Proceedings of ACM CoNEXT, pp. 15–26 (2010)Google Scholar
  15. 15.
    Guo, C., Wu, H., et al.: DCell: a scalable and fault-tolerant network structure for data centers. ACM SIGCOMM CCR 38(4), 75–86 (2008)CrossRefGoogle Scholar
  16. 16.
    Hopps, C.E.: Analysis of an equal-cost multi-path algorithm (2000)Google Scholar
  17. 17.
    Kabbani, A., Vamanan, B., et al.: FlowBender: flow-level adaptive routing for improved latency and throughput in datacenter networks. In: Proceedings of the 10th ACM International on Conference on emerging Networking Experiments and Technologies, pp. 149–160. ACM (2014)Google Scholar
  18. 18.
    Kandula, S., Katabi, D., et al.: Dynamic load balancing without packet reordering. ACM SIGCOMM CCR 37(2), 51–62 (2007)CrossRefGoogle Scholar
  19. 19.
    Kandula, S., Sengupta, S., et al.: The nature of data center traffic: measurements & analysis. In: Proceedings of ACM IMC, pp. 202–208 (2009)Google Scholar
  20. 20.
    Khanna, A., Zinky, J.: The revised arpanet routing metric. In: ACM SIGCOMM CCR, vol. 19, pp. 45–56 (1989)CrossRefGoogle Scholar
  21. 21.
    Luczak, M.J., McDiarmid, C., et al.: On the power of two choices: balls and bins in continuous time. Ann. Appl. Probab. 15(3), 1733–1764 (2005)MathSciNetCrossRefGoogle Scholar
  22. 22.
    Mascolo, S., Casetti, C., et al.: TCP westwood: bandwidth estimation for enhanced transport over wireless links. In: Proceedings of ACM MobiCom, pp. 287–297 (2001)Google Scholar
  23. 23.
    Mitzenmacher, M.: The power of two choices in randomized load balancing. IEEE Trans. Parallel Distrib. Syst. 12(10), 1094–1104 (2001)CrossRefGoogle Scholar
  24. 24.
    Raiciu, C., Barre, S., et al.: Improving datacenter performance and robustness with multipath TCP. In: ACM SIGCOMM CCR, vol. 41, pp. 266–277 (2011)CrossRefGoogle Scholar
  25. 25.
    Sen, S., Shue, D., et al.: Scalable, optimal flow routing in datacenters via local link balancing. In: Proceedings of ACM CoNEXT, pp. 151–162 (2013)Google Scholar
  26. 26.
    Shafiee, M., Ghaderi, J.: A simple congestion-aware algorithm for load balancing in datacenter networks. IEEE/ACM Trans. Netw. 25(6), 3670–3682 (2017)CrossRefGoogle Scholar
  27. 27.
    Vanini, E., Pan, R., Alizadeh, M., Taheri, P., Edsall, T.: Let it flow: resilient asymmetric load balancing with flowlet switching. In: Proceedings of NSDI. USENIX (2017)Google Scholar
  28. 28.
    Wischik, D., Raiciu, C., et al.: Design, implementation and evaluation of congestion control for multipath TCP. In: Proceedings of NSDI, vol. 11, p. 8 (2011)Google Scholar
  29. 29.
    Wu, X., Yang, X.: DARD: distributed adaptive routing for datacenter networks. In: Proceedings of IEEE ICDCS, pp. 32–41 (2012)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2018

Authors and Affiliations

  • Songyun Wang
    • 1
  • Xin Li
    • 1
    • 2
    Email author
  • Zhuzhong Qian
    • 2
  • Jiabin Yuan
    • 1
  1. 1.College of Computer Science and TechnologyNanjing University of Aeronautics and AstronauticsNanjingChina
  2. 2.State Key Laboratory for Novel Software TechnologyNanjing UniversityNanjingChina

Personalised recommendations