Performance analysis of multi-homed transport protocols with network failure tolerance Y. Qiao, E. Fallon, J. Murphy, L. Murphy, X. Zhu, G. Hayes, A. Matthews and A. Hanley Abstract: The performance of multi-homed transport protocols tolerant of network failure is studied. It evaluates the performance of different retransmission policies combined with path failure detection thresholds, infinite or finite receive buffers for various path bandwidths, delays and loss rate conditions through stream control transmission protocol simulation. The results show that retransmission policies perform differently with different path failure detection threshold configurations. It identifies that retransmission of all data on an alternate path with the path failure detection threshold set to zero performs the best in symmetric path conditions but its performance degrades acutely in asymmetric path conditions even when the alternate path delay is shorter than the primary path delay. It illustrates that retransmission of all data on the same path with the path failure detection threshold set to one or zero gives the most stable performance in all path configurations. 1 Introduction Multi-homing technologies, where a host can be addressed by multiple IP addresses, are increasingly being considered by developers implementing mobile applications. An enabling factor reinforcing this adoption is the trend towards mobile devices supporting a hybrid of networking capabilities such as 802.11 and universal mobile telecom- munications system (UMTS). The characteristics of mobile environments, with the possibility of frequent dis- connections and fluctuating bandwidth, pose significant issues for mobile application developers and therefore path redundancy offered by multi-homing protocols has a clear attraction. Furthermore, seamless handover between heterogeneous networks could be achieved by using multi- homing technologies [1–3]. This paper studies the performance of transport protocols, which support network failure tolerance through multi- homing technologies. The traditional transport layer proto- cols, such as transmission control protocol (TCP) and user datagram protocol (UDP), only support one IP address at each endpoint per connection. Therefore much effort has been put into the designing of multi-homing protocols to overcome the deficiencies of the traditional transport proto- cols. Currently, two multi-homing transport protocols have been proposed. They are stream control transmission proto- col (SCTP) [4] and datagram congestion control protocol (DCCP) [5]. DCCP is an unreliable transport protocol with congestion control, whereas SCTP is a reliable trans- port layer protocol and employs a similar congestion control mechanism to TCP. As this paper focuses on the performance of different retransmission policies and path failure detection thresholds in multi-homed environments, only SCTP is considered. The retransmission algorithms in SCTP consist of the fast retransmission algorithm and the timeout retransmission algorithm. According to path selection strategies during retransmissions, three retransmission policies have been proposed in [6], and are investigated in this paper. They are: † AllRtxAlt: All Retransmissions to an Alternate Path; † AllRtxSame: All Retransmissions to the Same Path; † FrSameRtoAlt: Fast Retransmissions to the Same Path, Timeout Retransmissions to an Alternate Path. AllRtxAlt is the default retransmission policy in base SCTP [4]. The authors of [6] evaluated these three retrans- mission policies with different extensions and the default SCTP parameters [such as Path.Max.Retrans (PMR) ¼ 5] in various lossy environments. The results show that FrSameRtoAlt with the multiple fast retransmission algor- ithm and the timestamp or the heartbeat after retransmission timeout (RTO) extension perform the best among the three policies and their respective extensions. AllRtxAlt performs the worst because of the stale RTO problem [6]. The problem for AllRtxSame is that the transmission time is very long when link failure occurs in the last data trans- mission round. SCTP detects path failure through protocol parameters PMR and RTO. When the number of consecutive trans- mission timeouts on a path exceeds PMR, SCTP will mark the path as inactive. In [7], the authors studied the per- formance of different PMR settings with FrSameRtoAlt and the multiple fast retransmission extension [6]. The results show that PMR ¼ 0 can achieve the best throughput in various path failure or non-failure situations. This paper studies the performance of these three retrans- mission policies with different PMR values in various sym- metric and asymmetric path conditions. Only bulky data transmission is considered. Real-time data transmission will be studied in future work. No extension is used # The Institution of Engineering and Technology 2008 doi:10.1049/iet-com:20070113 Paper first received 15th March and in revised form 14th October 2007 Y. Qiao, E. Fallon, X. Zhu, G. Hayes, A. Matthews and A. Hanley are with the Software Research Centre, Athlone Institute of Technology, Ireland J. Murphy and L. Murphy are with the Performance Engineering Laboratory, University College Dublin, Ireland Y. Qiao is also with the Institute of Software, Chinese Academy of Sciences, China and Graduate University of Chinese Academy of Sciences, China E-mail: ysqiao@ait.ie IET Commun., 2008, 2, (2), pp. 336–345 336