基于SCTP的接收端Data Reneging研究

发布时间：2018-02-14 18:13

本文关键词： SCTP Data Reneging检测模型发送缓存利用率吞吐量 SACK　出处：《东北大学》2014年硕士论文　论文类型：学位论文

【摘要】：数据接收端在接收到无序数据后将其存储在接收缓存,并发送选择确认SACK给数据发送端,但在该数据传送给接收应用进程之前将其从接收缓存中删除,这种现象称为Data Reneging 。现在的可靠传输协议TCP、SCTP都容忍Reneging的发生。RFC2018中描述：SACK选项是一个报告,即他通知数据发送端接收到该数据,数据接收端允许在发送选择确认之后删除该数据。本文主要研究了SCTP中Reneging发生的原因、影响以及在不同操作系统中检测Reneging 。由于TCP和SCTP的设计都容忍Reneging的发生,所以数据发送端必须保存发送数据的副本直到收到累计确认ACK后才可以将其从发送缓存中删除。如果Reneging在数据传输的过程中发生了,那么存储在发送缓存中的无序数据副本就用于数据重传,保证数据的可靠传输。相反,数据在传输的过程中没有发生Reneging,那么无序数据副本就没有必要保存,保存了不必要的数据造成了操作系统资源的浪费。SCTP的Reneging检测模型是在TCP中检测Reneging模型的基础上建立的。这个模型在SCTP的数据发送端检测Reneging 。通过比较接收缓存状态和新接收到的ACK的信息,当两者出现不一致时推断发生了Reneging 。根据接收到的SACK和ACK信息推断接收缓存的状态信息,但是在有些情况下接收到的SACK信息存在错误。有时候本应该发送的SACK信息没有发送,或者是发送了错误的SACK信息。Ekiz发现了7中错误的SACK错误行为,这7中错误行为中除第一种不会在SCTP中发生外,其他都可能发生在SCTP中。由于这些错误行为的存在导致该检测模型的准确性降低,所以在该模型中加入了对错误行为的分析。通过分析重传数据流,确定发生了Reneging还是错误行为引起的错误推断°通过对Reneging实例的研究,深入分析了Reneging产生的原因和影响。经研究发现TCP和SCTP的这种容忍Reneging的机制是不恰当的,这种机制导致发送缓存的利用率和系统吞吐量降低,为此有必要修改这种机制来提高发送缓存的利用率和系统的吞吐量。通过对SACK选项的语义进行修改或当检测到Reneging时重置(RESET)连接,可能会改善目前发送缓存利用率和吞吐量低的状况。
[Abstract]:The data receiver stores the unordered data in the receiving cache after receiving it, and sends the selected confirmation SACK to the data sender, but removes the data from the receiving cache before the data is transmitted to the receiving application process. This phenomenon is called Data Reneging. Today's reliable transport protocol, TCP / TCP, both tolerate the occurrence of Reneging. RFC2018 describes the: sack option as a report that he notifies the data sender of receiving the data. The data receiving side allows the data to be deleted after sending the selection confirmation. This paper mainly studies the reason why Reneging occurs in SCTP, its influence and the detection of Reneging in different operating systems. Since both TCP and SCTP are designed to tolerate the occurrence of Reneging, So the data sender must save a copy of the sent data until it receives the cumulative acknowledgement ACK before it can be removed from the sending cache. If the Reneging occurs during the data transfer, An unordered copy of the data stored in the sending cache is then used to retransmit the data to ensure the reliable transmission of the data. The Reneging detection model which saves unnecessary data causes the waste of operating system resources. SCTP is based on the detection of Reneging model in TCP. This model detects Reneging at the data sender of SCTP. State and newly received ACK messages, Reneging is inferred to occur when the two are inconsistent. The status of the received cache is inferred from the received SACK and ACK information, but in some cases there are errors in the SACK information received. Sometimes the SACK message that should have been sent is not sent. Or send the wrong SACK message. Ekiz found the wrong SACK error behavior in 7, except for the first one that doesn't happen in SCTP, The other may occur in SCTP. Because the accuracy of the detection model is reduced due to the existence of these error behaviors, an analysis of the error behavior is added to the model. To determine whether the Reneging occurred or the false inference caused by the wrong behavior the causes and effects of Reneging are deeply analyzed through the study of Reneging examples. It is found that the mechanism of TCP and SCTP's tolerance of Reneging is inappropriate. This mechanism leads to a decrease in the utilization of the sending cache and the throughput of the system, For this reason, it is necessary to modify this mechanism to increase the utilization of the sending cache and the throughput of the system by modifying the semantics of the SACK option or resetting the Reneging connection when the Reneging is detected. This may improve the current low send cache utilization and throughput.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.08

【相似文献】