Skip to main content

The Actual Cost of Programmable SmartNICs: Diving into the Existing Limits

  • Conference paper
  • First Online:
Book cover Advanced Information Networking and Applications (AINA 2021)

Abstract

Programmable Data Planes is a novel paradigm that enables efficient offloading of network applications. An important enabler for this paradigm is the current available SmartNICs, which should satisfy rigid network requirements such as high throughput and low latency. Despite recent research in this field, not much attention was given to understand the costs and limitations of offloading network applications into SmartNIC devices. Existing offloading approaches either neglect the existing limitations of SmartNICs or assume that as a fixed cost – leading, therefore, to sub-optimal offloading approaches. In this work, we conduct a comprehensive evaluation of SmartNICs in order to quantify existing performance limitations. We provide insights on network performance metrics such as throughput and packet latency while considering different key building blocks of complex P4 programs (e.g., registers, cryptography functions, or packet recirculation). Results show that line-rate throughput can degrade up to 8x, while latency can increase as much as 80x when performing memory-intensive operations in the data plane.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 169.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 219.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    https://www.dpdk.org/.

  2. 2.

    https://github.com/emmericp/MoonGen/tree/master/examples/netronome-packetgen.

  3. 3.

    https://github.com/mcluizelli/performanceSmartNIC.

References

  1. Ben Basat, R., Einziger, G., Friedman, R., Luizelli, M.C., Waisbard, E.: Constant time updates in hierarchical heavy hitters. In: Proceedings of the ACM SIGCOMM, SIGCOMM 2017, pp. 127–140. ACM, New York (2017)

    Google Scholar 

  2. Bosshart, P., Daly, D., Gibb, G., Izzard, M., McKeown, N., Rexford, J., Schlesinger, C., Talayco, D., Vahdat, A., Varghese, G., Walker, D.: P4: programming protocol-independent packet processors. ACM SIGCOMM 44(3), 87–95 (2014)

    Article  Google Scholar 

  3. Bosshart, P., Gibb, G., Kim, H.S., Varghese, G., McKeown, N., Izzard, M., Mujica, F., Horowitz, M.: Forwarding metamorphosis: fast programmable match-action processing in hardware for SDN. ACM SIGCOMM CCR 43(4), 99–110 (2013)

    Article  Google Scholar 

  4. Castro, A.G., Lorenzon, A.F., Rossi, F.D., Da Costa Filho, R.I.T., Ramos, F.M.V., Rothenberg, C.E., Luizelli, M.C.: Near-optimal probing planning for in-band network telemetry. IEEE Commun. Lett. 1 (2021). https://ieeexplore.ieee.org/document/9330755/keywords#keywords

  5. Emmerich, P., Gallenmüller, S., Raumer, D., Wohlfart, F., Carle, G.: MoonGen: a scriptable high-speed packet generator. In: Proceedings of the ACM IMC, IMC 2015, pp. 275–287. ACM, New York (2015)

    Google Scholar 

  6. Grant, S., Yelam, A., Bland, M., Snoeren, A.C.: SmartNIC performance isolation with FairNIC: programmable networking for the cloud. In: Proceedings of the ACM SIGCOMM, pp. 681–693 (2020)

    Google Scholar 

  7. Harkous, H., Jarschel, M., He, M., Priest, R., Kellerer, W.: Towards understanding the performance of p4 programmable hardware. In: ACM/IEEE Symposium on Architectures for Networking and Communications Systems, pp. 1–6. IEEE (2019)

    Google Scholar 

  8. Hohemberger, R., Castro, A.G., Vogt, F.G., Mansilha, R.B., Lorenzon, A.F., Rossi, F.D., Luizelli, M.C.: Orchestrating in-band data plane telemetry with machine learning. IEEE Commun. Lett. 23(12), 2247–2251 (2019)

    Article  Google Scholar 

  9. Hohemberger, R., Lorenzon, A.F., Rossi, F.D., Luizelli, M.C.: A heuristic approach for large-scale orchestration of the in-band data plane telemetry problem. In: Barolli, L., Amato, F., Moscato, F., Enokido, T., Takizawa, M. (eds.) Advanced Information Networking and Applications, pp. 381–392. Springer International Publishing (2020)

    Google Scholar 

  10. Liu, M., Cui, T., Schuh, H., Krishnamurthy, A., Peter, S., Gupta, K.: Offloading distributed applications onto SmartNICs using iPipe. In: Proceedings of the ACM Special Interest Group on Data Communication, pp. 318–333 (2019)

    Google Scholar 

  11. Netronome: Internet (2020). https://www.netronome.com/static/app/img/products/silicon-solutions/WP_NFP4000_TOO.pdf

  12. Pizzutti, M., Schaeffer-Filho, A.E.: Adaptive multipath routing based on hybrid data and control plane operation. In: IEEE INFOCOM, pp. 730–738 (2019)

    Google Scholar 

  13. Qiu, Y., Kang, Q., Liu, M., Chen, A.: Clara: performance clarity for SmartNIC offloading. In: Proceedings of the ACM Hot Topics in Networks, pp. 16–22 (2020)

    Google Scholar 

  14. Sanvito, D., Siracusano, G., Bifulco, R.: Can the network be the AI accelerator? In: Proceedings of the Workshop on In-Network Computing, pp. 20–25 (2018)

    Google Scholar 

  15. Sapio, A., Abdelaziz, I., Aldilaijan, A., Canini, M., Kalnis, P.: In-network computation is a dumb idea whose time has come. In: Proceedings of the 16th ACM Workshop on Hot Topics in Networks, pp. 150–156 (2017)

    Google Scholar 

  16. Siracusano, G., Bifulco, R.: In-network neural networks. arXiv preprint arXiv:1801.05731 (2018)

  17. Wang, S.Y., Wu, C.M., Lin, Y.B., Huang, C.C.: High-speed data-plane packet aggregation and disaggregation by p4 switches. J. Netw. Comput. Appl. 142, 98–110 (2019)

    Article  Google Scholar 

  18. Wang, S., Meng, Z., Sun, C., Wang, M., Xu, M., Bi, J., Yang, T., Huang, Q., Hu, H.: SmartChain: enabling high-performance service chain partition between SmartNIC and CPU. In: IEEE International Conference on Communications, pp. 1–7. IEEE (2020)

    Google Scholar 

  19. Xiong, Z., Zilberman, N.: Do switches dream of machine learning? Toward in-network classification. In: Proceedings of the 18th ACM Workshop on Hot Topics in Networks, pp. 25–33 (2019)

    Google Scholar 

Download references

Acknowledgements

This work was partially funded by National Council for Scientific and Technological Development (CNPq) (grant 427814/2018-9), São Paulo Research Foundation (FAPESP) (grant 2018/23092-1), Rio Grande do Sul Research Foundation (FAPERGS) (grants 19/2551-0001266-7,20/2551-000483-0, 19/2551-0001224-1).

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Marcelo C. Luizelli .

Editor information

Editors and Affiliations

Rights and permissions

Reprints and permissions

Copyright information

© 2021 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this paper

Check for updates. Verify currency and authenticity via CrossMark

Cite this paper

Viegas, P.B., de Castro, A.G., Lorenzon, A.F., Rossi, F.D., Luizelli, M.C. (2021). The Actual Cost of Programmable SmartNICs: Diving into the Existing Limits. In: Barolli, L., Woungang, I., Enokido, T. (eds) Advanced Information Networking and Applications. AINA 2021. Lecture Notes in Networks and Systems, vol 225. Springer, Cham. https://doi.org/10.1007/978-3-030-75100-5_17

Download citation

Publish with us

Policies and ethics