高可用度叢集(HA Cluster)，是一種針對關鍵性的服務保障其高可用性的叢集系統，其主要保護的對象，為提供關鍵服務的伺服器，稱為叢集節點。藉由備援軟硬體的提供，叢集節點可以組成一個群組。當群組中任何一個關鍵元件故障時，服務並不會就此中斷，而是經由故障轉移機制來啟動備援。這樣的備援機制可有效移除單點失敗，並降低系統停止提供服務的時間。
TCP傳輸控制協議，在網路上是最普遍，也是最常使用在關鍵服務的通訊協定。但是，TCP連線的行為特質並未被考慮在高可用度叢集的設計中。故障轉移期間，TCP封包將會漏失，而這些掉落的封包則被擁塞控制機制當成網路擁塞的訊號，並錯誤的啟動擁塞控制機制，降低傳輸速率。也使得故障轉移後，被中斷的連線無法立即重新開始傳輸。
在本篇論文中，我們提出一種近端回復的機制。此機制乃是藉由位於高可用度叢集兩端的負載平衡交換機來對通過其上的封包做選擇性近端快取，並在故障轉移完成後立即送出快取的封包，如此一來，即可達成有效率的近端回復機制。另外，對於雙作業模式的叢集，或是擁有不對稱鏈路頻寬的叢集，考量到故障轉移後可能發生頻寬不足的問題，我們接著提出了故障轉移後的速率控制機制，來保證錯誤轉移後TCP流量可持續穩定的傳輸。
模擬結果顯示，採用近端回復機制的叢集，能夠在故障轉移事件發生時，有效地避免浪費系統可用卻未被使用的時間。若進一步考慮故障轉移後，頻寬不足的情況，在加上速率控制元件之後，網路效能即可在僅受到輕微之影響的狀態下，達成故障轉移後TCP 連線之間公平的頻寬利用，較穩定的系統吞吐率，並有效降低位於瓶頸的緩衝區使用量。

[1] J. Wolfgang and T. Sven, "Analysis of internet backbone traffic and header anomalies observed," in Proceedings of the 7th ACM SIGCOMM conference on Internet measurement San Diego, California, USA: ACM, 2007.
[2] M. Dahlin, B. B. V. Chandra, L. Gao, and A. Nayate, "End-to-end WAN service availability," IEEE/ACM Transactions on Networking (TON), vol. 11, pp. 300-313, 2003.
[3] V. Jacobson, "Congestion avoidance and control," ACM SIGCOMM Computer Communication Review, vol. 25, pp. 157-187, 1995.
[4] P. Lewis, "A High Availability Clustering Solution," Linux Journal, vol. 1999, 1999.
[5] J. Padhye and S. Floyd, "Identifying the TCP Behavior of Web Servers," ACM SIGCOMM, 2001.
[6] V. Paxson, "Automated packet trace analysis of TCP implementations," Proceedings of the ACM SIGCOMM'97 conference on Applications, technologies, architectures, and protocols for computer communication, pp. 167-179, 1997.
[7] J. C. Hoe, "Improving the start-up behavior of a congestion control scheme for TCP," Applications, Technologies, Architectures, and Protocols for Computer Communication, pp. 270-280, 1996.
[8] F. Kevin and F. Sally, "Simulation-based comparisons of Tahoe, Reno and SACK TCP," SIGCOMM Comput. Commun. Rev., vol. 26, pp. 5-21, 1996.
[9] S. Floyd and T. Henderson, "The NewReno Modification to TCP’s Fast Recovery Algorithm," RFC 2582, April 1999, 1999.
[10] M. Mathis, J. Mahdavi, S. Floyd, and A. Romanow, "TCP Selective Acknowledgment Options," RFC 2018, October 1996, 1996.
[11] V. Paxson and M. Allman, "Computing TCP’s Retransmission Timer," 2000.
[12] J. H. Hu, G. Feng, and K. L. Yeung, "Hierarchical Cache Design for Enhancing TCP Over Heterogeneous Networks With Wired and Wireless Links," IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, vol. 2, p. 205, 2003.
[13] S. James and M. Glenford, "Link layer-based TCP optimisation for disconnecting networks," SIGCOMM Comput. Commun. Rev., vol. 33, pp. 31-42, 2003.
[14] M. Ji-Cheol and L. Byeong Gi, "Rate-adaptive snoop: a TCP enhancement scheme over rate-controlled lossy links," IEEE/ACM Trans. Netw., vol. 14, pp. 603-615, 2006.
[15] D. Martin, R. H. Thomas, and M. Jeff, "Experience with "link-up notification" over a mobile satellite link," SIGCOMM Comput. Commun. Rev., vol. 34, pp. 93-104, 2004.
[16] H. J. Byun and J. T. Lim, "Explicit window adaptation algorithm over TCP wireless networks," Communications, IEE Proceedings-, vol. 152, pp. 691-696, 2005.
[17] W. Huan-Yun, T. Shih-Chiang, and L. Ying-Dar, "Assessing and improving TCP rate shaping over edge gateways," Computers, IEEE Transactions on, vol. 53, pp. 259-275, 2004.
[18] A. James, O. Michel, and Y. M. Delfin, "A self-regulating TCP acknowledgment (ACK) pacing scheme," Int. J. Netw. Manag., vol. 12, pp. 145-163, 2002.
[19] A. James, O. Michel, Y. M. Delfin, and Y. Zhonghui, "Improving network service quality with explicit TCP window control," Int. J. Netw. Manag., vol. 11, pp. 169-188, 2001.
[20] L. Kalampoukas, A. Varma, and K. K. Ramakrishnan, "Explicit window adaptation: a method to enhance TCP performance," Networking, IEEE/ACM Transactions on, vol. 10, pp. 338-350, 2002.
[21] Y. Matsushita, T. Matsuda, and M. Yamamoto, "TCP Congestion Control with ACK-Pacing for Vertical Handover," IEICE TRANSACTIONS on Communications, vol. 90, pp. 885-893, 2007.
[22] K. Shrikrishna, K. Shivkumar, B. Prasad, and P. Bob, "TCP rate control," SIGCOMM Comput. Commun. Rev., vol. 30, pp. 45-58, 2000.
[23] "NLANR PMA Trace," http://pma.nlanr.net/.
[24] M. Mellia and H. Zhang, "TCP model for short lived flows," Communications Letters, IEEE, vol. 6, pp. 85-87, 2002.
[25] N. Cardwell, S. Savage, and T. Anderson, "Modeling TCP latency," in INFOCOM 2000. Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings. IEEE, 2000, pp. 1742-1751 vol.3.
[26] J. Padhye, V. Firoiu, D. F. Towsley, and J. F. Kurose, "Modeling TCP Reno performance: a simple model and its empiricalvalidation," Networking, IEEE/ACM Transactions on, vol. 8, pp. 133-145, 2000.
[27] W. R. Stevens, "TCP/IP illustrated (vol. 1): the protocols," 1993.
[28] "ns-2," http://www.isi.edu/nsnam/ns/.
[29] R. McBride, "Firewall Failover with pfsync and CARP."