CISCO IOS QOS TECHNOLOGY思科ios qos 技术
Classification and Marking分类和标记Packet classification features allow traffic to be partitioned into multiple priority levels, or classes of service. Packets can be classified in a variety of different ways-ranging from input interface, to NBAR for difficult to classify applications, to arbitrary access control lists. Classification is the first component of Modular QoS CLI (MQC), the simple, scalable, and powerful Cisco IOS QoS framework. MQC allows for clear separation of classification, from the policy applied on the classes, to the application of a QoS policy on an interface or sub-interface. Packets can also be marked in a variety of ways (ie: Layer2-802.1p/Q/ISL, ATM CLP bit, Frame-Relay DE-bit, MPLS EXP bits, etc., Layer3-IP Precedence, Differentiated Services Code Point (DSCP)) using the policy framework component of the MQC.数据包分类的特点是允许流量被分成多个优先级别，或是服务类别。数据包被非常广泛多种变化的分类从入接口，对于NBAR（基于网络应用层识别）去分类应用服务是困难的，必须加上访问控制列表。这种简单，可升级，强大的思科IOS QOS 框架中，QOS CLI（命令行接口）标（MQC）的第一个组件是类别。MQC允许清除类别不同，从被应用策略的分类在应用了QOS策略的接口或是子接口上。数据包能够被标记用多种办法当使用MQC策略框架组件（例如：2层的802.p/Q/ISL，ATM CLP bit, Frame-Relay DEbit, MPLS EXP bits, 等等其实就是常说COS, 3层的IP Precedence（优先级）,
Differentiated Services Code Point (DSCP)等等其实就是常说TOS） Congestion Avoidance 拥塞避免 The WRED algorithm provides for congestion avoidance on network interfaces by providing buffer management, and allowing Transmission Control Protocol (TCP) traffic to throttle back before buffers are exhausted. This helps avoid tail drops, and global synchronization issues, thereby maximizing network utilization and TCP-based application performance. The policy framework component of MQC accommodates WRED. 对于WRED（权重随机早期保护）算法提供拥塞避免在接口上通过提供缓冲区管理和允许TCP（传输控制协议）流量及时停止在发现缓冲区溢出。这就帮助避免的尾部丢弃，和解决全局同步，从而最优化网络和基于tcp的应用服务的性能。MQC策略框架组件就需要WRED Congestion Management 拥塞管理 Often a network interface is congested (even at high speeds, transient congestion is observed), and queuing techniques are necessary to ensure that the critical applications get the forwarding treatment necessary. For example, real time applications such as VoIP, stock-trading, etc. may need to be forwarded with the least latency and jitter (up to a provisioned limit). Cisco Low Latency Queuing (LLQ) provides this type of solution. For other non-delay sensitive traffic (such as File Transfer Program (FTP), HyperText Transfer Protocol (HTTP), etc.), other queuing techniques such as CBWFQ, and Modified Deficit Round Robin (MDRR) may be used. The queuing techniques can also be instantiated using the policy framework of the MQC. 队列技术可以确保关键应用程序得到优先转发，当网络借口经常被拥塞（甚至是快速，瞬间拥塞被发现）。例如RTP（时实应用程序）像VoIP，股票交易等等，都必须被转发以低延迟及抖动（或是规定限制边界）。思科的LLQ（低延迟队列）提供这种类型的解决办法。对于另外一些非敏感延迟流量（像文件传输协议（FTP），超文本传输协议（http）等等），可能使用另外一些队列技术例如CBWFQ（基于类的权重公平队列），和MDRR（改进不足轮询队列）。队列技术同样能够展示MQC策略框架组件 Traffic Conditioning 流量状态 Traffic entering a network can be conditioned by using a policer or shaper. A policer enforces a rate-limit, while a shaper limits the traffic flow to a specified rate using buffers. Mechanisms such as CAR, GTS, and FRTS can be configured without/within the MQC framework. 流量进入一个网络被优化通过监控和整形。当一个整形限制传输流量用特定速率通过缓冲区那么一个监控就加强速率的限制，这种机制例如CAR（用户指定速率），GTS（一般传输整形），和FRTS（帧中继传输整形）将被配置在MQC策略框架组件中 Signaling 打标识 In addition to supporting provisioned QoS (including the IETF DiffServ model with techniques such as CAR, GTS, L3 packet marking), Cisco IOS Software also provides for the IETF IntServ model. RSVP is the primary mechanism to perform admission control for flows in a network. An example can be found in VoIP. A call is completed only if the resources are available for it, ensuring that a call coming into a network does not bump or affect the quality of existing calls. Another technique called QoS Policy Propagation via BGP (QPPB) allows for indirectly signaling (using the community list attribute in BGP) the forwarding priority for packets destined toward an autonomous system, AS-path, or IP-prefix. This is a very useful feature for service providers and large enterprises. 为了能够支持规范的QOS（包括IETF（互联网工程任务组）的差分服务模型技术像CAR，GRS三层数据包标记）思科规范操作系统（交换机和路由器shell）也同样提供IETF的集成服务模型。在网络中RSVP（预留资源协议）是一个主要的机制去执行管理控制流量。例如一个VoIP的发现。一个呼叫被完成如果这个资源是预留的对于它，确保一个呼叫进入网络不冲击或影响已建立的呼叫质量。另外一个技术被称做QOS 策略移植通过BGP（QPPB）允许间接标识（使用共同体列出BGP路由属性）转发优先级的数据包到一个自治域，AS-path，or IP-prefix。这是一个非常有用的特点对于大型企业的服务提供。 Link Efficiency Mechanisms 连接有效机制 Streaming video and voice traffic uses the Real Time Protocol (RTP). IP, UDP, and RTP packet headers can be compressed from approximately 40 down to 5-8 bytes. This saves a tremendous amount of bandwidth in the case of low speed links, and when a large number of media streams are being supported. In addition, FRF.12 specification for frame fragmentation and Cisco Link Fragmentation and Interleaving (LFI) allow for fragmenting large data packets, interleaving them with RTP packets, and maintaining low delay and jitter for media streams. 视频和语音流量使用RTP，IP，UDP，and RTP包头来压缩从大约40个字节到5-8个字节。当许多流媒体被支持时，这样就节省了巨大的可用带宽在低速链路上。此外，FRF.12详细介绍了帧分裂传输和思科联结分裂和交叠技术（LFI）允许分割一个大的数据包，交叠他们使用RTP封装，以及维持一个低延迟和抖动的流媒体 Hardware Support 硬件支持 Cisco 800, 1700, 1800, 2600, 2800, 3600, 3700, 3800, 7200, 7300, 7500, 7600, 10000, and 12000 Series Routers 思科 800，1700，1800，2600, 2800, 3600, 3700, 3800, 7200, 7300, 7500, 7600, 10000, 和12000系列路由器 Cisco Catalyst 3500, 3700, 4000, and 6000 Series Switches 思科3500, 3700, 4000, and 6000系列交换机