Online Elephant Flow Prediction for Load Balancing in Programmable Switch-Based DCN
Pages 745 - 758
Abstract
In the data center network, traffic has a distinct heavy-tailed distribution characteristic, with the minority of throughput-sensitive elephant flows occupy most of the bandwidth and the majority of latency-sensitive mice flows require low latency. Therefore, it is very important to predict the network flow size and make a reasonable balanced scheduling. Currently, the traditional elephant flow detection schemes based on thresholds have poor accuracy and low granularity, while the intelligent detection schemes based on SDN has a certain flow scheduling response delay. For this reason, a two-stage online elephant flow prediction method for load balancing (OPLB) is proposed. Based on the programmable data plane, OPLB first pre-identifies the elephant flow by extracting the stateless features of the first packet of the network flow arriving at the switch. Secondly, the size of the elephant flow is predicted by extracting the features of the first <inline-formula> <tex-math notation="LaTeX">${n}$ </tex-math></inline-formula> packets of the flow. Finally, the detected elephant and mice flows are balanced to high throughput and low latency paths. Combined with the computing and storage capabilities of the programmable switch, the models and parameters in OPLB can be updated online by mapping the trained classification and prediction decision tree models to the matching-action pipelines of the programmable switch, thus achieving dynamic load balancing. We prototype OPLB in P4 software simulation environment and evaluate it with packet traces from the university data centers (UNI). The experiment shows that the accuracy of classification and prediction reached 89.3% when the proportion of elephant flow was 20%. At the same time, compared to the scheme that only uses single stage elephant flow prediction, OPLB reduces the amount of information collected by the switch by about 40% when the elephant flow proportion is 20%.
References
[1]
T. Chen, X. Gao, and G. Chen, “The features, hardware, and architectures of data center networks: A survey,” J. Parallel Distrib. Comput., vol. 96, pp. 45–74, Oct. 2016. 10.1016/j.jpdc.2016.05.009.
[2]
R. Rojas-Cessa, Y. Kaymak, and Z. Dong, “Schemes for fast transmission of flows in data center networks,” IEEE Commun. Surveys Tuts., vol. 17, no. 3, pp. 1391–1422, 3rd Quart., 2015. 10.1109/COMST.2015.2427199.
[3]
J. Zhang, F. R. Yu, S. Wang, T. Huang, Z. Liu, and Y. Liu, “Load balancing in data center networks: A survey,” IEEE Commun. Surveys Tuts., vol. 20, no. 3, pp. 2324–2352, 3rd Quart., 2018. 10.1109/COMST.2018.2816042.
[4]
C. E. Hopps. “Analysis of an equal-cost multi-path algorithm.” ETSI, Sophia Antipolis, France, RFC 2992, 2000. [Online]. Available: https://datatracker.ietf.org/doc/html/rfc2992
[5]
S. Kandula, S. Sengupta, A. Greenberg, P. Patel, and R. Chaiken, “The nature of data center traffic: Measurements & analysis,” in Proc. 9th ACM SIGCOMM Conf. Internet Meas., 2009, pp. 202–208.
[6]
T. Benson, A. Akella, and D. A. Maltz, “Network Traffic Characteristics of Data Centers in the Wild,” in Proc. 10th ACM SIGCOMM Internet Meas. Conf., 2010, pp. 267–280.
[7]
L. Liao, H.-C. Chao, and M.-Y. Chen, “Intelligently modeling, detecting, and scheduling elephant flows in software defined energy cloud: A survey,” J. Parallel Distrib. Comput., vol. 146, pp. 64–78, Dec. 2020. 10.1016/j.jpdc.2020.07.008.
[8]
M. Hamdanet al., “DPLBAnt: Improved load balancing technique based on detection and rerouting of elephant flows in software-defined networks,” Comput. Commun., vol. 180, pp. 315–327, Dec. 2021. 10.1016/j.comcom.2021.10.013.
[9]
Q. Shi, F. Wang, and D. Feng, “IntFlow: Integrating per-packet and perflowlet switching strategy for load balancing in datacenter networks,” IEEE Trans. Netw. Service Manag., vol. 17, no. 3, pp. 1377–1388, Sep. 2020.
[10]
J. Ye, C. Chen, and Y. H. Chu, “A weighted ECMP load balancing scheme for data centers using P4 Switches,” in Proc. IEEE 7th Int. Conf. Cloud Netw. (CloudNet), Piscataway, NJ, USA, 2018, pp. 1–4.
[11]
H. Zhang, J. Zhang, W. Bai, K. Chen, and M. Chowdhury, “Resilient datacenter load balancing in the wild,” in Proc. Conf. ACM Spec. Interest Group Data Commun., Los Angeles, CA, USA, 2017, pp. 253–266.
[12]
S. Ghorbani, Z. Yang, P. B. Godfrey, Y. Ganjali, and A. Firoozshahian, “DRILL: Micro load balancing for low-latency data center networks,” in Proc. Conf. ACM Special Interest Group Data Commun., Los Angeles, CA, USA, 2017, pp. 225–238.
[13]
M. Alizadehet al., “CONGA: Distributed congestion-aware load balancing for datacenters,” in Proc. Conf. ACM Special Group Data Commun., 2014, pp. 503–514.
[14]
K. He, E. Rozner, K. Agarwal, W. Felter, J. B. Carter, and A. Akella, “Presto: Edge-based load balancing for fast datacenter networks,” in Proc. ACM Conf. Special Interest Group Data Commun., 2015, London, U.K., 2015, pp. 465–478.
[15]
L. Jing, L. Jie, G. Shou, Y. Hu, and D. Wei, “SDN based load balancing mechanism for elephant flow in data center networks,” in Proc. Int. Symp. Wireless Pers. Multimedia Commun., 2015, pp. 486–490.
[16]
F. Tang, H. Zhang, L. T. Yang, and L. Chen, “Elephant flow detection and load-balanced routing with efficient sampling and classification,” IEEE Trans. Cloud Comput., vol. 9, no. 3, pp. 1022–1036, Jul.-Sep. 2021. 10.1109/TCC.2019.2901669.
[17]
A. R. Curtis, W. Kim, and P. Yalagandula, “Mahout: Low-overhead datacenter traffic management using end-host-based elephant detection,” in Proc. 30th IEEE Int. Conf. Comput. Commun. Joint Conf. IEEE Comput. Commun. Soc., Shanghai, China, 2011, pp. 1629–1637.
[18]
S. C. Madanapalli, M. Lyu, H. Kumar, H. H. Gharakheili, and V. Sivaraman, “Real-time detection, isolation and monitoring of elephant flows using commodity SDN system,” in Proc. IEEE/IFIP Netw. Operat. Manag. Symp., 2018, pp. 1–5.
[19]
A. R. Curtis, J. C. Mogul, J. Tourrilhes, P. Yalagandula, P. Sharma, and S. Banerjee, “DevoFlow: Scaling flow management for highperformance networks,” in Proc. ACM SIGCOMM Conf. Appl. Technol. Architect. Protocol. Comput. Commun., 2011, pp. 254–265.
[20]
B. Wang, J. Su, J. Li, and B. Han, “EffiView: Trigger-based monitoring approach with low cost in SDN,” in Proc. 19th IEEE Int. Conf. High Perform. Comput. Commun., 2017, pp. 309–315.
[21]
A. AlGhadhban and B. Shihada, “FLight: A fast and lightweight elephant-flow detection mechanism,” in Proc. 38th IEEE Int. Conf. Distrib. Comput. Syst., 2018, pp. 1537–1538.
[22]
H. Li, H. Lu, and X. Fu, “An optimal and dynamic elephant flow scheduling for SDN-based data center networks,” J. Intell. Fuzzy Syst., vol. 38, no. 1, pp. 247–255, 2020. 10.3233/JIFS-179399.
[23]
Q. Tang, H. Zhang, J. Dong, and L. Zhang, “Elephant flow detection mechanism in SDN-based data center networks,” Sci. Program., vol. 2020, pp. 1–9, Sep. 2020. 10.1155/2020/8888375.
[24]
L. Che and B. Qiu, “Landmark LRU: An efficient scheme for the detection of elephant flows at Internet routers,” IEEE Commun. Lett., vol. 10, no. 7, pp. 567–569, Jul. 2006. 10.1109/LCOM.2006.224421.
[25]
H. Xu, H. Huang, S. Chen, and G. Zhao, “Scalable software-defined networking through hybrid switching,” in Proc. IEEE Conf. Comput. Commun., 2017, pp. 1–9.
[26]
P. R. Torres Jr., A. García-Martínez, M. Bagnulo, and E. P. Ribeiro, “An elephant in the room: Using sampling for detecting heavy-hitters in programmable switches,” IEEE Access, vol. 9, pp. 94122–94131, 2021. 10.1109/ACCESS.2021.3092281.
[27]
M. V. B. D. Silva, A. S. Jacobs, R. J. Pfitscher, and L. Z. Granville, “IDEAFIX: Identifying elephant flows in P4-based IXP networks,” in Proc. IEEE Global Commun. Conf., 2018, pp. 1–6.
[28]
M. Al-Fares, S. Radhakrishnan, B. Raghavan, N. Huang, and A. Vahdat, “Hedera: Dynamic flow scheduling for data center networks,” in Proc. 7th USENIX Symp. Netw. Syst. Design Implement. (NSDI), Berkeley, CA, USA, 2010, pp. 89–92.
[29]
H. Zhang, F. Tang, and L. Barolli, “Efficient flow detection and scheduling for SDN-based big data centers,” J. Ambient Intell. Humaniz. Comput., vol. 10, no. 5, pp. 1915–1926, 2019. 10.1007/s12652-018-0783-6.
[30]
P. Jurkiewicz, “Boundaries of flow table usage reduction algorithms based on elephant flow detection,” in Proc. IFIP Netw. Conf., 2021, pp. 1–9.
[31]
C. Bi, X. Luo, T. Ye, and Y. Jin, “On precision and scalability of elephant flow detection in data center with SDN,” in Proc. IEEE Globecom Workshops (GC Wkshps), 2013, pp. 1227–1232.
[32]
S.-C. Chao, K. C.-J. Lin, and M.-S. Chen, “Flow classification for software-defined data centers using stream mining,” IEEE Trans. Services Comput., vol. 12, no. 1, pp. 105–116, Jan./Feb. 2019. 10.1109/TSC.2016.2597846.
[33]
Z. Xiong and N. Zilberman, “Do switches dream of machine learning?: Toward in-network classification,” in Proc. 18th ACM Workshop Hot Topics Netw., Princeton, NJ, USA, Nov. 2019, pp. 25–33.
[34]
J. H. Lee and K. Singh, “SwitchTree: In-network computing and traffic analyses with random forests,” Neural Comput. Appl., vol. 2020, pp. 1–12, Nov. 2020. [Online]. Available: https://doi.org/10.1007/s00521-020-05440-2
[35]
B. M. Xavier, R. S. Guimares, G. Comarela, and M. Martinello, “Programmable Switches for in-Networking Classification,” in Proc. IEEE Conf. Comput. Commun., 2021, pp. 1–10.
[36]
C. Kim, A. Sivaraman, N. Katta, A. Bas, A. Dixit, and L. J. Wobker, “Inband network telemetry via programmable dataplanes,” in Proc. Conf. ACM Special Interest Group Data Commun. (SIGCOMM), New York, NY, USA, 2015, pp. 1–2.
[37]
X. Zhang, L. Cui, F. P. Tso, and W. Jia, “pHeavy: Predicting heavy flows in the programmable data plane,” IEEE Trans. Netw. Service Manag., vol. 18, no. 4, pp. 4353–4364, Dec. 2021. 10.1109/TNSM.2021.3094514.
[38]
Y. Li, J. Gao, E. Zhai, M. Liu, K. Liu, and H. H. Liu, “Cetus: Releasing P4 Programmers from the chore of trial and error compiling,” in Proc. 19th USENIX Symp. Netw. Syst. Design Implement., 2022, pp. 371–385.
[39]
W. Wang, Y. Sun, K. Zheng, M. A. Kâafar, D. Li, and Z. Li, “Freeway: Adaptively isolating the elephant and mice flows on different transmission paths,” in Proc. 22nd IEEE Int. Conf. Netw. Protocols (ICNP), Piscataway, NJ, USA, 2014, pp. 362–367.
[40]
M. Al-Fares, A. Loukissas, and A. Vahdat, “A scalable, commodity data center network architecture,” in Proc. Conf. ACM Special Interest Group Data Commun. (SIGCOMM), New York, NY, USA, 2008, pp. 63–74.
[41]
C. Sammut and G. I. Webb, Encyclopedia of Machine Learning and Data Mining. New York, NY, USA: Springer, 2017. 10.1007/978-1-4899-7687-1.
[42]
X. Peng, Q. Wenyu, Q. Heng, X. Yujie, and L. Zhiyang, “An Efficient Elephant Flow Detection with Cost-sensitive in SDN,” in Proc. 1st Int. Conf. Ind. Netw. Intell. Syst. (INISCom), 2015, pp. 24–28.
[43]
L. Breiman, J. H. Friedman, R. A. Olshen, and C. J. Stone, Classification and Regression Trees. New York, NY, USA: Routledge, 1984.
[44]
S. Xie, G. Hu, X. Wang, C. Xing, and Y. Liu, “A decision treebased online traffic classification method for QoS routing in data center networks,” Security Commun. Netw., vol. 2022, Nov. 2022, Art. no. 10.1155/2022/9419676.
[45]
“Mininet: Rapid prototyping for software defined networks.” 2021. [Online]. Available: https://github.com/mininet/mininet
[46]
“P4 Software switch-behavioral model version 2 (BMV2).” Accessed: Dec. 12, 2021. [Online]. Available: https://github.com/p4lang/behavioral-model
Index Terms
- Online Elephant Flow Prediction for Load Balancing in Programmable Switch-Based DCN
Index terms have been assigned to the content through auto-classification.
Recommendations
Flow-based packet-mode load-balancing for parallel packet switches
Parallel Packet Switch (PPS) is broadly used in designing large-capacity switching fabrics, since it resolves the bottlenecks of scheduling algorithm, memory bandwidth, and serial transmission technology simultaneously. Current packet-based load-...
Elephant Flow Detection Mechanism in SDN-Based Data Center Networks
Aiming at the problem of network congestion and unbalanced load caused by a large amount of data capacity carried in elephant flow in the data center network, an elephant flow detection method based on SDN is proposed. This method adopts the autodetect ...
Autonomic load balancing of flow monitors
In monitoring flows at routers for flow analysis or deep packet inspection, the monitor calculates hash values from the flow ID of each packet arriving at the input port of the router. Therefore, the monitors must update the flow table at the ...
Comments
Information & Contributors
Information
Published In
1932-4537 © 2023 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See https://www.ieee.org/publications/rights/index.html for more information.
Publisher
IEEE Press
Publication History
Published: 25 September 2023
Qualifiers
- Research-article
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 0Total Downloads
- Downloads (Last 12 months)0
- Downloads (Last 6 weeks)0
Reflects downloads up to 13 Nov 2024
Other Metrics
Citations
View Options
View options
Get Access
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Sign in