네트워크 트래픽 상태에 적응적인 UDT Rate Control 기법

An Adaptive UDT Rate Control Method on Network Traffic Condition

  • 투고 : 2010.12.02
  • 심사 : 2011.03.17
  • 발행 : 2011.03.25

초록

광 네트워크 기술의 발달로 인해 오늘날 대부분의 네트워크는 수십 Gb/s를 지원한다. UDT(UDP-based Data Transfer protocol)는 대역폭과 지연시간이 큰 네트워크 환경에서 대용량 데이터 전송을 위한 프로토콜이다. 그러나 UDT는 고정된 sync-interval(0.01초) 간격으로 주기적인 rate control을 실행한다. 네트워크 트래픽 상태의 변화가 심한 환경에서는 sync-interval 동안 가용 대역폭을 충분히 활용하지 못하는 문제점이 발생한다. 본 논문은 sync-interval 구간에서 네트워크 트래픽 상태에 적응적인 rate control 기법을 제안한다. 네트워크 트래픽 상태는 RTT 변화를 바탕으로 판단한다. 제안 기법은 네트워크 상태를 4개의 구간으로 설정하여 패킷 sending period를 조절함으로써 rate control을 수행한다. 시뮬레이션 결과를 통하여 기존 UDT에 비해 throughput 과 안정성이 향상되었음을 보였다.

Due to the growth of optical fiber network technologies, most networks recently support several tens of Gb/s bandwidth. UDT(UDP-based Data Transfer protocol) is an optimized protocol for bulk data transmission on the network, which has the high bandwidth and long delay time. It periodically controls the sending rate on the fixed sync-interval, 0.01 sec. Here, if the network traffic status varies quickly, available network bandwidth is not able to be properly utilized in-between the sync-intervals. In this paper, we propose an adaptive rate control method with considering the network traffic status in-between the sync-interval periods. The network traffic status is determined based on the RTT variances. With dividing the network status into four classes, the proposed method performs a new rate control by adjusting the inter-packet sending period for a corresponding class. As a simulation result, the proposed method improves the message delivery throughput as well as stability than that of the existing UDT Protocol.

키워드

참고문헌

  1. A. Chien, T. Faber, A. Falk, J. Bannister, R. Grossman, and J. Leigh, "Transport Protocols for High Performance: Whither TCP," Communications of the ACM, vol. 46, no. 11, pp. 42-49, Mar. 2003. https://doi.org/10.1145/948383.948408
  2. W. Feng and P. Tinnakornsrisuphap, "The Failure of TCP in High-performance Computational Grids," Proceedings on IEEE conference on Supercomputing '00, pp. 4-10, Nov. 2000.
  3. 박태준, 이재용, 김병철, "대역폭과 지연의 곱이 큰 네트워크를 위한 개선된 TCP 혼잡제어 메카니즘," 전자공학회논문지, 42권 TC편, 1호, 126-134쪽, 2006년.
  4. L. Xu, K. Harfoush, and I. Rhee, "Binary Increase Congestion Control(BIC) for Fast Long-Distance Networks," Proceedings on IEEE INFOCOM '04, pp. 2514-2524, Mar. 2004.
  5. I. Rhee and L. Xu, "CUBIC: A New TCP-Friendly High-Speed TCP Variant," Proceedings on Third International Workshop on Protocols for Fast Long-Distance Networks, Feb. 2005.
  6. C. Jin, D. X. Wei, and S. H. Low, "FAST TCP : Motivation, Architecture, Algorithms, Performance," Proceedings on IEEE INFOCOM 2004, pp. 2490-2501, Mar. 2004.
  7. E. He, J. Leigh, O. Yu, and T. A. Defanti, "Reliable Blast UDP : Predictable High Performance Bulk Data Transfer," Proceedings on International Conference on IEEE'02, pp. 317-324, Sep. 2002.
  8. M. Meiss, "Tsunami: A High-Speed Rate-Controlled Protocol for File Transfer," www.evl.uic.edu/eric/atpTSUNAMI.pdf, 2009.
  9. Y. Gu and R. L. Grossman, "SABUL: A Transport Protocol for Grid Computing," Grid Computing, vol. 1, no. 4, pp. 377-386, Dec. 2003. https://doi.org/10.1023/B:GRID.0000037553.18581.3b
  10. Y. Gu and R. L. Grossman. "UDT: UDP-based Data Transfer for High Speed Wide Area Networks," Computer Networks, vol. 51, no. 7, pp. 1777-1799, May 2007. https://doi.org/10.1016/j.comnet.2006.11.009
  11. Y. Gu and R. L. Grossman, "UDTv4: Improvements on Performance and Usability," Gridnets 2008, vol. 2, no. 1, pp. 9-23, Oct. 2008.
  12. Y. Gu, X. Hong, and R. L. Grossman, "Experiences in Design and Implementation of a High Performance Transport Protocol," Proceedings on ACM SC'04, pp. 6-12, Nov. 2004.
  13. J. Postel, "Transmission Control Protocol," IETF, RFC 793, Sep. 1981.
  14. M. Allman, V. Paxson, and W. Stevens, "TCP Congestion Control," IETF, RFC 2581, Apr. 1999.
  15. 안도식, 왕기철, 조기환, "UDT환경에서 RTT예측에 의한 Sync-Interval 구간의 Rate Control 기법," 한국정보처리학회 추계학술대회, 17권, 2호, pp. 1038-1041, Nov. 2010.
  16. Network Simulator, ns-2, http://www.isi.edu/nsnam/ns