数据中心网络中的MPTCP Incast（英文）

In recent years,dual-homed topologies have appeared in data centers in order to offer higher aggregate bandwidth by using multiple paths simultaneously.Multipath TCP(MPTCP) has been proposed as a replacement for TCP in those topologies as it can efficiently offer improved throughput and better fairness.However,we have found that MPTCP has a problem in terms of incast collapse where the receiver suffers a drastic goodput drop when it simultaneously requests data over multiple servers.In this paper,we investigate why the goodput collapses even if MPTCP is able to actively relieve hot spots.In order to address the problem,we propose an equally-weighted congestion control algorithm for MPTCP,namely EW-MPTCP,without need for centralized control,additional infrastructure and a hardware upgrade.In our scheme,in addition to the coupled congestion control performed on each subflow of an MPTCP connection,we allow each subflow to perform an additional congestion control operation by weighting the congestion window in reverse proportion to the number of servers.The goal is to mitigate incast collapse by allowing multiple MPTCP subflows to compete fairly with a single-TCP flow at the shared bottleneck.The simulation results show that our solution mitigates the incast problem and noticeably improves goodput in data centers.     

Modeling TCP Incast Issue in Data Center Networks and an Adaptive Application-Layer Solution

In data centers, the transmission control protocol (TCP) incast causes catastrophic goodput degradation to applications with a many-to-one traffic pattern. In this paper, we intend to tame incast at the receiver-side application. Towards this goal, we first develop an analytical model that formulates the incast probability as a function of connection variables and network environment settings. We combine the model with the optimization theory and derive some insights into minimizing the incast probability through tuning connection variables related to applications. Then, enlightened by the analytical results, we propose an adaptive application-layer solution to the TCP incast. The solution equally allocates advertised windows to concurrent connections, and dynamically adapts the number of concurrent connections to the varying conditions. Simulation results show that our solution consistently eludes incast and achieves high goodput in various scenarios including the ones with multiple bottleneck links and background TCP traffic.     

A survey on TCP Incast in data center networks

In high bandwidth, low latency data center networks, when multiple data senders simultaneously communicate with a single receiver, namely in many‐to‐one communication pattern, the burst data overload the receiver's switch buffers, which leads to Transmission Control Protocol (TCP) throughput collapse. This is the so‐called TCP Incast problem. It has become a new hot research topic in recent years. Also, many proposals have been put forward from the aspects of multiple layers including link layer, transport layer, and application layer, etc., to mitigate TCP Incast. In this paper, an in‐depth survey on these proposals is given, and the principles, merits, and drawbacks of the main proposals for solving the TCP Incast issue are analyzed and summarized. Copyright © 2012 John Wiley & Sons, Ltd.     

一种适合数据链的TOP拥塞控制策略

传统因特网传输控制协议（TCP）主要针对有线网络设计,无法应对数据链网络中突发干扰、高误码率、拓扑高动态变化引起的拥塞和吞吐率下降等问题。提出一种适合数据链网络的拥塞控制策略,利用主动探测结果来设定拥塞窗口大小,采用反馈机制进行动态拥塞控制,通过网关节点的缓存和转发来实现快速重传。仿真结果表明,与典型拥塞控制机制相比,提出的策略可明显改善数据链网络的传输性能。     

软件定义数据中心网络研究

提出了一种基于软件定义网络的数据中心网络方案,并对其结构和部署进行了详细介绍。通过使用软件定义网络技术,该方案解决了传统数据中心中网络利用效率低下、管控方式复杂、建设运维成本较高等问题。该方案具有结构简单、易于部署的优点。在负载均衡、域内虚拟机迁移等典型应用中,我们对该方案进行了验证,结果表明该方案具有良好的实用性。该方案为数据中心网络架构设计提供了一种重要思路。     

TCP SkyLine:数据中心网络高吞吐率传输

针对数据中心网络在"多对一"并发流量模式下,TCP（Transmission Control Protocol）及其现有改进方案在单轮数据传输和多轮数据传输下吞吐率低下问题,提出了一种通过数据包标记实现丢包快速发现和快速重传并动态调整拥塞窗口初始值的策略,称为TSL（TCP SkyLine）.TSL同时解决了传统TCP Incast问题和多轮数据传输下由遗留窗口引发的TCP Incast问题.实验表明,TSL在单轮数据传输和多轮数据传输下均能获得90%以上的带宽利用率.在10Gbps网络中,其支持的并发连接数与传统TCP和DCTCP相比分别提升了5倍和1倍,有效吞吐率分别提升了18倍和8.6倍;在1Gbps网路中,支持的并发连接数较传统TCP和DCTCP分别提升了5.8倍和1倍.     

模块化数据中心机房的实现方式

本文针对数据中心机房提出模块化设计理念,可以有效解决现有数据中心的一些问题,使数据中心具备灵活性、可扩展性、高标准、绿色节能、并能降低人力成本。本文重点介绍了模块化数据中心的实现方式。     

A Simulation Study of Scalable TCP and HighSpeed TCP in Geostationary Satellite Networks

This paper investigates the performance of two TCP enhancements (i.e., Scalable TCP and HighSpeed TCP), recently proposed for high-speed backbone networks with a very large bandwidth-delay product, in a geostationary satellite environment. Both persistent and elastic traffic patterns are considered, performance being evaluated in terms of TCP throughput and mean session delay, respectively. The impact of channel characteristics (packet error rate, correlation between successive losses) is widely discussed. Fairness issues are also accounted for, together with the impact of some system parameters, such as the satellite link bandwidth. Extensive comparisons are carried out among Scalable TCP, HighSpeed TCP and other congestion control schemes. The obtained results show the soundness for the use of such protocols in geostationary satellite networks. Indeed, both protocols permit to improve the performance of TCP connections in a wide range of channel conditions, showing (especially Scalable TCP) to be able to cope well with rainy conditions and to exploit a future increase in the satellite link capacity. This work was carried out within the framework of the SatNex Network of Excellence, http://www.satnex.org     

一种基于链路传输延迟的MPTCP拥塞控制算法*

由于传统TCP拥塞控制算法直接应用到MPTCP(Multipath Transport Control Protocol)中存在公平性问题,以及不能有效地发挥多路径传输的优势,因而从公平性方面对MPTCP现有拥塞控制算法进行研究.研究发现,现有的MPTCP拥塞控制算法均受到相同的回路时间限制.提出一种基于链路延迟的RTT补偿算法(Compensating for RTT mismatch,C-RTT ).该算法通过设置网络带宽占用比参数以及对MPTCP连接的子流设置侵略因子,从而保证瓶颈链路处MPTCP数据流和TCP数据流公平地共享可用带宽.最后通过NS3仿真实验证明,该算法能够有效地补偿链路中因RTT不等引起的公平性问题,并避免链路之间数据的非周期抖动,且保证了多路径传输的优越性.     

卫星IP网络中传输控制协议的研究

传输控制协议(TCP)是TCP / IP协议栈的主要传输协议,它在网络中的传输效率是影响IP网络传输性能的关键因素之一。当网络中存在卫星链路时,卫星链路的一些固有特性会使TCP的传输性能受到一定影响。文中首先指出卫星信道的特点,对TCP协议在卫星网络中的传输性能进行了分析,重点介绍最近几年这方面的研究成果。     

光突发交换(OBS)网络TCP性能改善

介绍光突发的基本原理及影响OBS网络TCP性能的一些因素，并针对TCP性能的改善的两种方法进行了详细说明，提出了一种改进策略。     

EVA: a better TCP version for resource‐insufficient networks

TCP Vegas exhibits unfair congestion avoidance mechanism, which aggravates when there are insufficient network resources to accommodate buffer space of a pipe (bandwidth delay product). To remedy this shortcoming, we propose an Enhanced VegAs (EVA) that employs three auxiliary mechanisms: Δ revision, congestion detection and congestion tendency detection. A 2^k factorial design with replications is used to study the effect of the three mechanisms. Our results show that TCP EVA achieves better performance than Vegas under various network conditions. Furthermore, congestion avoidance schemes, such as TCP EVA, perform much better than congestion control schemes, such as TCP Reno, in resource‐insufficient networks. Copyright © 2002 John Wiley & Sons, Ltd.     

一种无线传感器网络公平数据收集方案(英文)

To solve the slow congestion detection and rate convergence problems in the existing rate control based fair data collection schemes, a new fair data collection scheme is proposed, which is named the improved scheme with fairness or ISWF for short. In ISWF, a quick congestion detection method, which combines the queue length with traffic changes of a node, is used to solve the slow congestion detection problem, and a new solution, which adjusts the rate of sending data of a node by monitoring the channel utilization rate, is used to solve the slow convergence problem. At the same time, the probability selection method is used in ISWF to achieve the fairness of channel bandwidth utilization. Experiment and simulation results show that ISWF can effectively reduce the reaction time in detecting congestion and shorten the rate convergence process. Compared with the existing tree-based fair data collection schemes, ISWF can achieve better fairness in data collection and reduce the transmission delay effectively, and at the same time, it can increase the average network throughput by 9.1% or more.     

基于深度增强学习的数据中心网络coflow调度机制

最小化语义相关流的平均完成时间是数据中心网络流量管理面临的难题之一.受人工智能领域深度增强学习方向的最新研究进展启发,本文提出一种的新的语义相关流调度机制.将带宽约束的语义相关流调度问题转化为连续的学习过程,通过学习以往策略实现最佳调度.引入反向填充和有限复用机制,保证系统的工作保持性和无饥饿性.仿真结果表明,在不同的网络负载下,本文提出的调度机制均使得语义相关流的平均完成时间小于其他调度机制,尤其是网络负载较大时,相比最先进的调度机制,性能提升约50%.     

TCP/IP网络的动态模型描述

拥塞现象成为TCP/IP网络发展面临的一个重要问题。因此，拥塞控制对TCP/IP网络的鲁棒性和稳定性具有重要作用。目前，网络拥塞控制策略主要包括两类：端到端的控制机制，如TCP拥塞控制算法；网络内部的主动队列管理(AQM)策略。但由于缺乏对网络系统动态特性的了解，这些拥塞控制策略大都基于专家经验，并没有建立完整的理论分析框架。为此，本文从数据流的角度出发，通过建立网络基本单元状态方程模型来实现IP网络系统的数学解析模型，然后用混杂系统来描述TCP带有拥塞控制策略的数据传输过程，建立了TCP/IP网络的动态模型，为网络系统中动态性能的分析、拥塞控制策略的设计奠定了基础。实验结果表明，该数学模型与NS仿真实验的结果相一致。     

基于松弛时间与累计发送量的数据中心网络混合流调度机制

数据中心网络中同时存在截止时间流（deadline flow）和非截止时间流（non-deadline flow）,为降低非截止时间流的平均完成时间（Average Flow Complete Time,AFCT）同时维持低截止时间错失率（Deadline Miss Rate,DMR）,本文提出了一种基于松弛时间与累计发送量的混合流调度机制（Slack Time and Accumulation based Mix-flow Scheduling,STAM）.首先通过引入松弛时间的概念,衡量截止时间流对非截止时间流在传输时延上的宽容度;然后根据松弛时间,通过使截止时间流尽可能接近其规定截止时间完成,降低非截止时间流的完成时间;最后,利用最小累计发送量优先策略进一步降低非截止时间流的平均完成时间.仿真结果表明,该机制能有效降低非截止时间流的平均完成时间,同时保证较低的截止时间错失率.     

TCP拥塞控制算法研究

随着各种网络应用的不断涌现,网络拥塞问题变得越来越严重,解决这一问题也就越来越重要。阐述了TCP拥塞控制的基本原理,然后介绍了现有的几种TCP拥塞控制算法,并对这些算法进行了分析比较,结果表明虽然每种算法都有其优点,但也存在不足,不能从根本上解决拥塞问题。文章最后指出了拥塞控制的问题所在以及解决方法的发展趋势。     

TCP拥塞控制机制及性能分析

随着因特网的迅猛发展,因特网中的业务量呈逐年上升趋势,研究了TCP拥塞控制机制及其提供的拥塞控制性能,特别是对TCP连接过程中的数据业务流量与丢失率和其他参数的关系进行了深入的研究。分析表明TCP拥塞控制机制可通过控制注入网络的业务流量,在一定程度下缓解网络拥塞程度,但不能实现数据的有效率的传输。     

无线自组织网络TCP容量优化

为了计算无线自组织网络中准确的TCP容量,并利用其进行有效的窗口控制,以避免拥塞窗口增长过速的问题,提出一种适用于无线自组织网络TCP容量的优化计算。此方法基于时延和空间复用率,考虑网络中竞争干扰,解决了以往的计算方法对拓扑不能广泛适用的问题。模拟结果显示,基于文章的TCP容量进行拥塞窗口调节,能有效地提高网络的吞吐量。     

TCP/IP协议的拥塞控制策略

网络拥塞问题会导致网络性能下降,对网络正常运行有很大影响,严重时会导致死锁的产生.文中分析了网络拥塞的产生的原因,并着重讨论了传统端到端的TCP拥塞控制的策略和方法,传统TCP拥塞控制策略提出了改进策略和网络层的IP拥塞控制策略.最后对基于TCP和IP拥塞控制策略进行了对比探讨.