Ming Li
ABSTRACT
When a TCP connection experiences a timeout, the sender must wait at least RTOmin (Minimum Retransmission Timeout) before doing the retransmission, during which the channel may be completely idle, undermining the throughput and channel efficiency. In this paper, we investigate the origin of RTOmin and find that it is needed to mitigate against spurious timeouts when the Delayed ACK (DA) scheme for TCP is implemented.
Motivated by this observation, we propose a deployable and TCP-compatible new Delayed ACK (NDA) to replace the legacy DA. Our solution differs with previous work is that instead of using complex algorithms or fine-grained timer to tune RTOmin, we modify the DA scheme with minor changes to allow the sender to remove the RTOmin constraint while reserving the delayed ACK function at the receiver. In order to eliminate the aggressiveness of RTO (Retransmission Timeout) after removing RTOmin, we use coding techniques to encode the timeout retransmitted packets to make the potential spurious retransmissions useful.
The simulation results demonstrate that in lossy wireless networks, NDA is efficient, because it achieves much higher TCP goodput and channel efficiency compared to DA. The gain we obtain by the use of NDA comes from two-order effects. One effect comes from removing the RTOmin constraint, because small RTO timer makes TCP react quickly to timeouts, resulting in small transmission idle. The other effect comes from eliminating consecutive RTO by allowing the receiver to acknowledge each timeout retransmission, which further reduces the RTO idle.
- Network simulator (ns-3) {Online} Available: http://www.nsnam.org.Google Scholar
- M. Alizadeh, A. Greenberg, D. A. Maltz, J. Padhye, P. Patel, B. Prabhakar, S. Sengupta, and M. Sridharan. Data center TCP (DCTCP). ACM SIGCOMM Computer Communication Review, 40(4):63--74, 2010. Google ScholarDigital Library
- M. Allman. Retransmission Timeout Considerations. Internet draft, May 2012.Google Scholar
- M. Allman and V. Paxson. On estimating end-to-end network path properties. In ACM SIGCOMM Computer Communication Review, volume 29, pages 263--274. ACM, 1999. Google ScholarDigital Library
- R. Braden. Requirements for Internet hosts - communication layers. IETF RFC 1122, 1989. Google ScholarDigital Library
- Y. Chen, R. Griffith, J. Liu, R. H. Katz, and A. D. Joseph. Understanding TCP incast throughput collapse in datacenter networks. In Proceedings of the 1st ACM workshop on Research on enterprise networking, pages 73--82. ACM, 2009. Google ScholarDigital Library
- T. Das and K. M. Sivalingam. TCP improvements for data center networks. In 2013 Fifth International Conference on Communication Systems and Networks (COMSNETS), pages 1--10. IEEE, 2013.Google ScholarCross Ref
- P. Devkota and A. N. Reddy. Performance of quantized congestion notification in TCP incast scenarios of data centers. In 2010 IEEE International Symposium on Modeling, Analysis & Simulation of Computer and Telecommunication Systems (MASCOTS), pages 235--243. IEEE, 2010. Google ScholarDigital Library
- A. Gurtov and R. Ludwig. Responding to spurious timeouts in TCP. In INFOCOM 2003. Twenty-Second Annual Joint Conference of the IEEE Computer and Communications. IEEE Societies, volume 3, pages 2312--2322. IEEE, 2003.Google ScholarCross Ref
- T. Henderson, S. Floyd, A. Gurtov, and Y. Nishida. The NewReno modification to TCP's fast recovery algorithm. RFC6582, 2012.Google Scholar
- T. Ho, M. Médard, R. Koetter, D. R. Karger, M. Effros, J. Shi, and B. Leong. A random linear network coding approach to multicast. IEEE Transactions on Information Theory, 52(10):4413--4430, 2006. Google ScholarDigital Library
- S. Katti, H. Rahul, W. Hu, D. Katabi, M. Médard, and J. Crowcroft. XORs in the air: practical wireless network coding. IEEE/ACM Transactions on Networking (TON), 16(3):497--510, 2008. Google ScholarDigital Library
- S. Kulkarni and P. Agrawal. A probabilistic approach to address TCP incast in data center networks. In 2011 31st International Conference on Distributed Computing Systems Workshops (ICDCSW), pages 26--33. IEEE, 2011. Google ScholarDigital Library
- J. Kumar Sundararajan, D. Shah, and M. Médard. ARQ for network coding. In IEEE International Symposium on Information Theory. ISIT 2008, pages 1651--1655. IEEE, 2008.Google ScholarCross Ref
- R. Ludwig and R. H. Katz. The Eifel algorithm: making TCP robust against spurious retransmissions. ACM SIGCOMM Computer Communication Review, 30(1):30--36, 2000. Google ScholarDigital Library
- V. Paxson, M. Allman, J. Chu, and M. Sargent. Computing TCP's retransmission timer. IETF RFC 6298, June 2011.Google Scholar
- I. Psaras and V. Tsaoussidis. On the properties of an adaptive TCP Minimum RTO. Computer Communications, 32(5):888--895, 2009. Google ScholarDigital Library
- A. Shpiner, I. Keslassy, G. Bracha, E. Dagan, O. Iny, and E. Soha. A switch-based approach to throughput collapse and starvation in data centers. Computer Networks, 2012. Google ScholarDigital Library
- J. K. Sundararajan, D. Shah, M. Médard, S. Jakubczak, M. Mitzenmacher, and J. Barros. Network coding meets TCP: Theory and implementation. Proceedings of the IEEE, 99(3):490--512, 2011.Google ScholarCross Ref
- V. Vasudevan, A. Phanishayee, H. Shah, E. Krevat, D. G. Andersen, G. R. Ganger, G. A. Gibson, and B. Mueller. Safe and effective fine-grained TCP retransmissions for datacenter communication. In ACM SIGCOMM Computer Communication Review, volume 39, pages 303--314. ACM, 2009. Google ScholarDigital Library
- H. Wu, Z. Feng, C. Guo, and Y. Zhang. ICTCP: Incast Congestion Control for TCP in data center networks. In Proceedings of the 6th International COnference, page 13. ACM, 2010. Google ScholarDigital Library
- Y. Zhang and N. Ansari. On mitigating TCP incast in data center networks. In Proceedings IEEE INFOCOM 2011, pages 51--55. IEEE, 2011.Google ScholarCross Ref
Index Terms
- Efficient new delayed ACK for TCP: old problem, new insight
Recommendations
An Adaptive Delayed Acknowledgment Strategy to Improve TCP Performance in Multi-hop Wireless Networks
In multi-hop wireless networks, transmission control protocol (TCP) suffers from performance deterioration due to poor wireless channel characteristics. Earlier studies have shown that the small TCP acknowledgments consume as much wireless resources as ...
On making TCP robust against spurious retransmissions
Despite the fact that Transmission Control Protocol (TCP) is a stable and mature protocol, and has been well tuned over years, changes in its congestion control are still in progress without altering the fundamental underlying dynamics of TCP congestion ...
Packet loss estimation of TCP flows based on the delayed ACK mechanism
APNOMS'09: Proceedings of the 12th Asia-Pacific network operations and management conference on Management enabling the future internet for changing business and new computing servicesBased on the Delayed ACK mechanism in TCP protocol, a new method is proposed to estimate the packet loss based on the variation of the rate between the bidirectional TCP packet numbers. This method made it possible to estimate the TCP packet loss based ...
Comments