WO2008011778A1 - Procédé et système de garantie de qualité de bout en bout - Google Patents

Procédé et système de garantie de qualité de bout en bout Download PDF

Info

Publication number
WO2008011778A1
WO2008011778A1 PCT/CN2007/001187 CN2007001187W WO2008011778A1 WO 2008011778 A1 WO2008011778 A1 WO 2008011778A1 CN 2007001187 W CN2007001187 W CN 2007001187W WO 2008011778 A1 WO2008011778 A1 WO 2008011778A1
Authority
WO
WIPO (PCT)
Prior art keywords
message
path
guaranteed
gateway
label
Prior art date
Application number
PCT/CN2007/001187
Other languages
English (en)
French (fr)
Inventor
Shihui Hu
Original Assignee
Huawei Technologies Co., Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huawei Technologies Co., Ltd filed Critical Huawei Technologies Co., Ltd
Publication of WO2008011778A1 publication Critical patent/WO2008011778A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/70Admission control; Resource allocation
    • H04L47/72Admission control; Resource allocation using reservation actions during connection setup
    • H04L47/724Admission control; Resource allocation using reservation actions during connection setup at intermediate nodes, e.g. resource reservation protocol [RSVP]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/10Flow control; Congestion control
    • H04L47/15Flow control; Congestion control in relation to multipoint traffic
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/70Admission control; Resource allocation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/70Admission control; Resource allocation
    • H04L47/82Miscellaneous aspects
    • H04L47/825Involving tunnels, e.g. MPLS

Definitions

  • the present invention relates to the field of network communication technologies, and in particular, to an end-to-end quality assurance method and system. Background technique
  • RSVP Resource Reservation Protocol
  • NSS Next Steps in Signaling
  • RSVP protocol based on multi-protocol label switching (MPLS, Multiprotocal Label Switching) traffic, so as to achieve end-to-end quality assurance.
  • MPLS Multiprotocal Label Switching
  • the ingress gateway When the data is forwarded, the ingress gateway performs the functions of the traffic classification, the policy, and the P-speed.
  • the packet is encapsulated into the corresponding MPLS traffic engineering tunnel for forwarding according to the resource reservation.
  • the resource reservation based on edge-to-edge flow aggregation is well implemented, and flow-based admission control and resource reservation ensure end-to-end performance, that is, , Specify link paths for edge-to-edge flow aggregation in the region and allocate bandwidth on each link.
  • end-to-end performance that is, Specify link paths for edge-to-edge flow aggregation in the region and allocate bandwidth on each link.
  • the resources of the ingress gateway to the egress gateway need to be allocated in advance, that is, if an ingress gateway to the egress gateway applies for the bandwidth resource of a link, other edge-to-edge links cannot apply for the bandwidth resource again. That is, the bandwidth resources of the links in the area are not shared by the edge-to-edge resources.
  • the embodiment of the invention provides an end-to-end quality assurance method.
  • an end-to-end service spans one or several network segments
  • the end-to-end quality is achieved by implementing edge-to-edge quality assurance in the network segment based on congestion warning. Guarantee.
  • An embodiment of the present invention provides an end-to-end quality assurance method, including:
  • An embodiment of the present invention provides an end-to-end quality assurance system, where the system includes: an ingress gateway, configured to determine whether an incoming message is a guaranteed message, and according to the resource when the incoming message is a guaranteed message The reservation policy determines a forwarding path, and adds an MPLS label stack and a label label to the guaranteed message;
  • An intermediate router configured to transmit the guaranteed message forwarded by the ingress gateway part to the egress gateway part, and mark and count the forwarding events experienced by the guaranteed message;
  • An egress gateway configured to identify the protected tag according to the tag, remove the tag tag and the MPLS label, and forward to the next hop;
  • the egress gateway generates resource occupation information based on the statistical parameters of the MPLS path of the ingress gateway to the egress gateway and the total statistical parameter value, and sends the resource information to the portal to enable the ingress gateway to perform resource management.
  • Management includes resource reservation and/or admission control.
  • the forwarded event information that is experienced is carried in the guaranteed message and the corresponding identifier is set, for example, on each egress link of the intermediate router, Knowing to monitor the congestion status of the link, and counting the forwarding event information and the information of the four residences for resource reservation and/or admission control, the invention can implement dynamic stream-based resource reservation, so that the edge is Edge flows can share resources on the link, improving link utilization.
  • the method and corresponding system for implementing resource reservation and admission control based on congestion warning provided by the embodiment of the present invention do not require complicated manual intervention, and the capability of network resource reservation is dynamically changed with the actual carrying capacity of the network segment. Change, thus improving the robustness of the network and the timeliness of admission control.
  • Figure 1 is a simplified model diagram illustrating a network segment
  • FIG. 2 is a flowchart of operations of an ingress gateway in a network segment according to an embodiment of the present invention
  • FIG. 3 is a flowchart of operations of an intermediate router in a network segment according to an embodiment of the present invention
  • FIG. 4 is a flowchart of operations of an egress gateway in a network segment according to an embodiment of the present invention
  • FIG. 5 is a flowchart of resource admission control according to an embodiment of the present invention.
  • 6 is a flag label structure of an MLPS path according to an embodiment of the present invention.
  • FIG. 8 is a system block diagram of an embodiment of the present invention. detailed description
  • the end-to-end service flow may span multiple network segments, and achieve end-to-end performance guarantee by implementing edge-to-edge performance guarantee on each segment of the path, in the entire end-to-end path. running a resource reservation protocol (such as RSVP or the NSIS etc.), the entire network as a resource reservation protocol intermediate node P
  • a resource reservation protocol such as RSVP or the NSIS etc.
  • Figure 1 shows a simple model diagram of a network segment. There is an end-to-end flow between user A and user B. User A to user B only pass through one network segment.
  • the devices on the network segment include the following functions: the ingress gateway, the intermediate router, and the egress gateway.
  • the ingress gateway to the egress gateway can establish one or more MPLS paths according to the network topology, which can be established during network initialization, or can be triggered when admission control is performed for the flow.
  • the operation process of the ingress gateway in the network segment of the embodiment of the present invention includes: Step S11: The incoming packet is classified into a non-guaranteed packet and a guaranteed packet according to the traffic classification; if the packet is a non-guaranteed packet Then normal forwarding; if it is a guaranteed message, step S12 is performed;
  • Step S12 performing corresponding operations according to a predetermined policy of the flow, such as speed limit, shaping, and marking operations;
  • Step S13 determining a forwarding path, that is, adding an MPLS label stack in front of the packet to specify a forwarding path of the packet, adding a flag label before the label stack, and forwarding the guaranteed packet to the intermediate router, where the label label indicates The message is a guaranteed message for use by the intermediate router.
  • Step S21 Determine, according to the flag label of the forwarded packet, whether it is a guaranteed packet, and retain the label for use by the egress gateway;
  • the path indicated by the MPLS label is forwarded to the egress gateway;
  • Step S22 Perform statistics on the egress queue according to the egress link. If the number of leaky bucket bytes used for statistics exceeds a threshold, the congestion flag is set proportionally. If it is a dominant congestion alarm mode, you need to calculate the flow rate based on the egress link. If the flow rate exceeds the threshold, you need to set the resource occupation flag for the packet.
  • Each link of the network segment is configured with a fixed bandwidth, or only a fixed bandwidth is configured for the link through which the guaranteed message passes, for transmitting the insured message.
  • a leaky bucket is used to monitor the congestion state of the link, and when there is an insured i ⁇ message, the leaky bucket is correspondingly increased.
  • the number of bytes, and according to the leaky bucket algorithm, the number of bytes of the leaky bucket is reduced at the configured rate, thereby providing a rate of control data injection into the network, smoothing the burst traffic mechanism on the network, and the burst traffic can be shaped to be a network.
  • the ingress gateway device also has the function of an intermediate router on the inner side of the connected network segment, and also needs to perform congestion monitoring based on the egress link.
  • two thresholds, a low value and a high value are set for the number of leaky bucket bytes of the link.
  • the guaranteed packets are forwarded normally.
  • the intermediate router sets the congestion flag for the guaranteed packets according to a certain ratio.
  • the ratio of setting the congestion flag can be linearly increased between the low threshold and the high threshold;
  • the probability of setting the congestion flag is 100%. For example, when the threshold is low, the probability of setting the congestion flag is 0%, the middle is n% (n ranges from 1 to 100), and the high threshold is 100%.
  • the triggering of the resource occupation alarm is roughly divided into two steps: The router in the network segment sends an occupation alarm to the egress gateway; the egress gateway sends an occupation alarm to the relevant ingress gateway.
  • Explicit trigger Set the traffic rate threshold on each link based on the egress link of the device. If the guaranteed flow rate exceeds the threshold, set the resource occupation alarm flag for the packets exceeding the threshold.
  • the egress gateway receives the resource occupation. After the packet of the alarm flag is sent, the occupation alarm message is sent to the relevant ingress gateway.
  • Implicit triggering The egress gateway detects the estimated congestion level of each path from side to side. When the congestion level evaluation value is in the preset trigger range, the egress gateway sends a resource occupation alarm message to the corresponding ingress gateway.
  • the scope can be specified by each operator according to the operation of its own system.
  • Step S31 Determine, according to the flag label, whether the received packet is a guaranteed packet, and if it is a non-guaranteed packet, if it is normally forwarded, if If the message is guaranteed, step S32 is performed;
  • Step S32 Calculate the congestion level of the entire ingress gateway to the egress gateway based on the MPLS path statistics flow rate and congestion level;
  • step S33 it is determined whether the alarm mode is recessive or dominant. If it is in the dominant mode, step S34 is performed, and if it is in recessive mode, step S35 is performed;
  • Step S34 determining whether there is an occupation resource alarm flag, if not, proceeding to step S37 to forward the message to the next hop, if there is a resource occupation alarm flag, proceeding to step S36;
  • Step S35 when the alarm mode is recessive mode, it is determined whether the congestion level exceeds the threshold, if not, then proceeds to step S37, otherwise in step S36;
  • Step S36 Send a resource occupation alarm message to the ingress gateway.
  • Step S37 Forward the packet to the next hop.
  • the guaranteed flag label and the MPLS label are removed before forwarding to the next hop.
  • Step S41 Determine the ingress gateway:
  • the egress gateway saves the information of the guaranteed message, according to the message marked with the resource occupation flag. Step S42, measuring the data flow rate of the guaranteed message:
  • the egress gateway measures a flow rate of the guaranteed message flow on each path from the ingress gateway to the egress gateway and a total flow rate of the guaranteed message from the ingress gateway to the egress gateway;
  • Step S43 Send a resource occupation alarm to the ingress gateway:
  • the egress gateway On the egress link of the device, set the traffic rate threshold on each link. If the guaranteed flow exceeds the threshold, set the resource occupation alarm flag to the packet exceeding the threshold, and the egress gateway receives the alarm flag marked with the resource occupation. After the packet is sent, an occupation alarm message is sent to the relevant ingress gateway.
  • the egress gateway detects the estimated congestion level of each path from side to side. When the congestion level evaluation value is in the preset trigger range, the egress gateway sends a resource occupation alarm message to the corresponding ingress gateway. The egress gateway sends a resource occupation alarm message to the ingress gateway, the message including an acceptable rate and congestion level evaluation value on each path from the ingress gateway to the egress gateway, and an edge-to-edge total rate and total congestion level evaluation value.
  • the ingress gateway After receiving the resource occupation alarm, the ingress gateway measures the traffic of the corresponding path of the resource occupation alarm, and also increases the measurement of other paths from the same edge to the edge. When the ingress flow rate on a path measured by the ingress gateway is greater than the flow rate measured by the egress gateway, it indicates that a packet loss occurs on the edge-to-edge path, and the ingress gateway needs to reduce the traffic. There are two methods. If the congestion level evaluation value is lower on other paths from the same edge to the edge, then part of the flow on the path is split to other paths, and if the edge to the edge is on all paths. If there are no remaining resources available, some services are suspended according to the business strategy.
  • the resource admission control process in the embodiment of the present invention includes the following steps:
  • the ingress gateway receives the RS VP PATH message, and the ingress gateway determines the egress gateway, and transparently passes the RS VP PATH message through the network segment to the egress gateway. And save the relevant information in the resource request message.
  • the egress gateway adds its own IP address to the RS VP PATH packet, and then forwards it to the next hop.
  • the egress gateway When receiving the RSVP RESV message from the downstream, the egress gateway sends the RSVP reservation message to the ingress gateway, and further carries the congestion level and the flow rate of each path from the ingress gateway to the egress gateway, and the total congestion level of the ingress gateway to the egress gateway. And flow rate;
  • the ingress gateway reserves information according to RSVP, the flow rate and congestion of each path on the path, and the like. Level information, determine the admission control of the resource reservation and which path the message passes. And storing information about the guaranteed message at the ingress gateway.
  • FIG. 6 is a diagram showing the structure of a flag tag added before a label stack of a packet forwarded on an MPLS path according to an embodiment of the present invention, which is 32 bits in total.
  • FIG. 7 is a flowchart showing the forwarding process of the label in the network segment from the ingress router to the egress router in the embodiment of the present invention.
  • an MPLS forwarding path based on the end-to-end flow aggregation; adding a 32-bit flag label to the label stack for the packet to be transmitted on the MPLS path, and the first 20 bits of the label label are used for the label value.
  • the value ranges from 0 to 15 for the label reservation value of the MPLS protocol.
  • the remaining 12 bits are used to identify the forwarding events that the packet experiences on the path.
  • the entry label edge router of the MPLS path completes the flow-based admission, classification, and The policy processing, encapsulating the corresponding MPLS forwarding label stack, and inserting a label label before the label stack; when the text experiencing a forwarding event such as congestion or data flow rate overrun, the label switching router on the MPLS path performs corresponding on the label label.
  • the exit label edge router of the path monitors the status of the path according to the forwarding status of each message identifier. The monitored path status may trigger or be used as an entry label edge router for admission and resource management parameters.
  • the status of the monitoring path includes dividing the number of bits of the message receiving the congestion flag within a period of time by the number of bits of the total message received during the period to obtain the congestion level of the path.
  • the end-to-end quality assurance system based on congestion warning mainly includes an ingress gateway part, an intermediate router part and an egress gateway part.
  • the ingress gateway section 100 includes: an admission control apparatus 101, a message classification judging apparatus 102, and a forwarding control apparatus 103.
  • the admission control device 101 participates in the aforementioned admission control process in conjunction with FIG.
  • the message classification judging device 102 judges whether the message is a guaranteed message or a non-guaranteed message, and if it is a guaranteed message, it is normally forwarded, otherwise the forwarding control device 103 operates according to the set policy, such as a speed limit, a flag. Then, first, add an MPLS label stack in front of the text to specify the forwarding path of the packet, and then add a flag label (the flag label indicates that the packet is a guaranteed message for the intermediate router) before the label stack. Forward.
  • the intermediate router portion 200 includes a message classification judging device 201, a forwarding control device 202, and a congestion monitoring device 203.
  • the message classification judging device 201 judges whether the message is a guaranteed message or a non-guaranteed message, and if it is a non-guaranteed message, it is normally forwarded, otherwise the forwarding control device 202 passes The path indicated by the MPLS label forwards the packet to the egress gateway; the congestion monitoring device 203 monitors the egress queue based on the egress link, and if the number of leaky buckets for monitoring exceeds the threshold, press The ratio sets the congestion flag for the message. If it is a dominant alarm mode, the intermediate router part further includes an occupation resource alarm device 204, configured to set a resource alarm flag for the message exceeding the threshold.
  • the device 302 counts the congestion level of the entire ingress gateway to the egress gateway based on the MPLS path statistics flow rate and the statistical congestion level. Then, the occupant resource alarm generating device 303 sends an occupant resource alarm message to the ingress gateway or directly forwards to the next hop according to different alarm modes according to the different resource alarm mode. The specific operation is as described above in conjunction with FIG.
  • the operator can implement dynamic resource allocation according to real-time services, fully utilize limited network resources, and also embody the package. Exchange network resource sharing features.

Description

一种端到端的质量保证方法及系统 本申请要求于 2006 年 07 月 20 日提交中国专利局、 申请号为 200610061724.2、 发明名称为 "基于拥塞预警的端到端的质量保证方法及系 统" 的中国专利申请的优先权, 其全部内容通过引用结合在本申请中。 技术领域
本发明涉及网络通信技术领域 , 尤其涉及一种端到端的质量保证方法 及系统。 背景技术
在当前的网络中, 端到端的质量保证是目前研究的一个热点课题。 目 前实现端到端的资源预留协议有资源预留协议 ( RSVP , Resource Reservation Protocol )和即将完成的下一代信令协议 ( NSIS, Next Steps in Signaling )。 RSVP不适用于大型的网络, 但是从 RSVP的视角可以把一个网段(实现差 分服务的网络或网段)看成一个 RSVP使能的网络节点, 这样在端到端的 路径上实现 RSVP协议就变得可能了。
目前针对端到端性能的保证方法很多, 其中一种比较好的解决方案是 基于多协议标签交换 ( MPLS, Multiprotocal Label Switching )流量工程运行 RSVP协议, 从而实现端到端的质量保证。 该方法的工作原理大致如下:
1、 在网段内部建立从入口网关到出口网关(边到边)的基于 MPLS的 流量工程, 并为入口网关到出口网关预留相应的资源, 预留的带宽是基于 入口网关到出口网关的流聚合的带宽;
2、 在入口网关和出口网关感知 RSVP协议消息, 进行基于统计的准入 控制和资源预留;
3、 在数据转发时, 入口网关完成流分类、 策略、 P艮速等功能, 根据流 申请资源预留的情况, 将报文封装到相应的 MPLS流量工程隧道进行转发。
基于流量工程的 RSVP 资源预留的方法, 很好地实现了基于边到边的 流聚合的资源预留, 以及基于流的准入控制和资源预留, 保证了端到端的 性能,也就是说, 在区域内为边到边的流聚合指定链路路径并在每个链路上 分配带宽。 但是上述方法中, 因为入口网关到出口网关的资源需要预先的分配, 即如果一个入口网关到出口网关申请了某链路的带宽资源, 则其它边到边 的链路将不能再申请该带宽资源, 也就是说, 区域内链路的带宽资源不会 被各边到边的资源预留共享。 当一个区域的网关很多时, 需要链路上有足 够的带宽资源用于分配。 具体来说, 如果一个网段有 n个网关, 如果要实 现所有边到边的资源预留的话, 共需要为 Nx ( N-1 ) 12个路径预留资源, .当该网段需要再增加一个网关时, 就得再增加 N个路径的资源预留, 这对 资源是个极大的浪费。 发明内容
本发明实施例提供一种端到端的质量保证方法, 当一个端到端的业务 横跨一个或数个网段, 基于拥塞预警通过实现网段内的边到边的质量保证 来实现端到端的质量保证。
本发明实施例提供一种实现端到端的质量保证方法, 包括:
在受保证报文中承载所经历的转发事件信息并设置相应的标识; 统计所述转发事件信息并^ ^据所统计的信息对所述受保证报文进行资源预 留和 /或准入控制。
本发明实施例提供一种实现端到端的质量保证系统, 该系统包括: 入口网关, 用于判断进入报文是否为受保证报文时, 并在该进入报文 为受保证报文时根据资源预留策略确定转发路径, 并为所述受保证报文添 加 MPLS标签栈和标志标签;
中间路由器, 用于将入口网关部分转发来的受保证报文传输到出口网 关部分, 并对所述受保证报文所经历的转发事件进行标记和统计; 和
出口网关, 用于才艮据所述标志标签识别受保证 ·ί艮文, 去掉所述标志标 签和 MPLS标签, 转发到下一跳;
所述出口网关基于入口网关到出口网关的 MPLS路径的统计参数以及 总的统计参数值产生资源占用信息, 并将所述资源信息发送到入口^关, 以使入口网关进行资源管理, 所 ¾资源管理包括资源预留和 /或准入控制。
本发明实施例提供的技术方案中 , 在受保证报文中承载所经历的转发 事件信息并设置相应的标识, 如在中间路由器的每个出口链路上, 设置标 识来监测该链路的拥塞状态 , 统计所述转发事件信息并 4居所统计的信息 进行资源预留和 /或准入控制, 采用本发明能实现动态的基于流的资源预 留, 使得边到边的流可以共享链路上的资源, 从而提高链路利用率。 本发 明实施例提供的基于拥塞预警来实现资源预留和准入控制的方法和相应的 系统, 不需要复杂的人工干预, 而且网络资源预留的能力随网段的实际运 载能力的改变而动态改变, 因此可提高网络的健壮性和准入控制的时效性。 附图说明
图 1是说明一个网段的简单的模型图;
图 2是本发明实施例的网段内的入口网关的操作流程图;
图 3是本发明实施例的网段内的中间路由器的操作流程图;
图 4是本发明实施例的网段内的出口网关的操作流程图;
图 5 是本发明实施例的资源准入控制流程图;
图 6是本发明实施例的 MLPS路径的标志标签结构;
图 7是本发明实施例的标志标签的转发路径;
图 8是本发明实施例的系统方框图。 具体实施方式
下面将结合附图具体说明本发明提供的基于拥塞预警的端到端质量保 证方法及相应的系统。 在此, 端到端的业务流可能横跨多个网段, 通过实 现所经路径上的每个网段内边到边的性能保证, 从而实现整个端到端的性 能保证, 在整个端到端路径上运行资源预留协议(如 RSVP或 NSIS等), 整个网段作为资源预留协议的一个中间节点 P
为了便于描述, 图 1给出的是一个网段的简单的模型图。 用户 A和用 户 B之间存在一条端到端的流, 用户 A到用户 B只经过一个网段。 如图 1 所示, 网段的设备从完成的功能来分包括: 入口网关、 中间路由器和出口 网关。 入口网关到出口网关可以根据网络拓朴建立一条或多条 MPLS路径, 可以在网络初始化时建立, 也可以在为流做准入控制时触发建立。
下面, 将结合附图分别说明根据本发明的基于拥塞预警的边到边质量 保证方法中入口网关、 中间路由器和出口网关的工作流程, 并说明网段内 准入控制的过程。 参见图 2, 本发明实施例的网段内的入口网关的操作流程包括: 步骤 S11 , 根据流分类将进入报文区分为非受保证报文和受保证报文; 若为非受保证报文则正常转发; 若为受保证报文, 则执行步骤 S12;
步骤 S12, 根据该流的预定策略进行相应的操作, 如限速、 整形和标志 等操作;
步骤 S13 ,确定转发路径, 即先在报文前面添加 MPLS标签栈指定该报 文的转发路径, 再在标签栈之前添加一个标志标签, 并将受保证报文转发 到中间路由器, 该标志标签标明该报文为受保证报文, 供中间路由器使用。
参见图 3, 发明实施例的网段内的中间路由器的操作流程如下: 步骤 S21 , 根据转发来的报文的标志标签判断其是否为受保证报文, 并 保留该标志标签供出口网关使用; 还经过 MPLS标签表明的路径将报文转 发到出口网关;
步骤 S22 , 进行基于出口链路对报文在出口队列的情况进行统计, 如 果用于统计的漏桶字节数超过阈值, 则按比例给艮文设置拥塞标志。 如果 是显性拥塞告警模式, 还需基于出口链路作流速统计, 如果流速超过阈值, 还需给报文设置资源占用标志。
网段的每一个链路都配置一个固定的带宽,或只为受保证报文所经过的 链路配置一个固定的带宽, 用以传送受保 i^艮文。
在本发明的实施例中, 在中间路由器的每个出口链路上, 利用一个漏桶 来监测该链路的拥塞状态,,当有受保 i^艮文到来时,相应地增加漏桶的字节 数, 而按照漏桶算法, 漏桶的字节数按配置的速率减少, 从而提供控制数据 注入到网络的速率, 平滑网络上的突发流量机制, 突发流量可以被整形以便 为网络提供一个稳定的流量。 通过漏桶即可得知相应链路上是否发生拥塞。 需要注意的是,入口网关设备在连接网段内侧也具有中间路由器的功能,也 需做基于出口链路的拥塞监测。
优选地, 对链路的漏桶字节数设两个阈值, 一个低阔值, 一个高阔值。 当漏桶字节数低于低阈值时, 受保证报文正常转发; 当漏桶字节数介于低 阈值和高阈值之间时, 中间路由器按一定比例将受保证报文设置拥塞标志, 设置拥塞标志的比例在低阈值和高阈值之间可以是线性增长的; 当漏桶字 节数高于高阈值时, 设置拥塞标志的几率为 100%。 例如低阈值时设置拥塞 标志的几率为 0 % , 中间为 n % ( n取值范围为 1 ~ 100 ) , 高阀值为 100 %。
触发资源占用告警大致分两个步骤: 网段内的路由器向出口网关发送 占用告警; 出口网关向相关的入口网关发送占用告警。 告警的触发有两种 模式即显性触发和隐性触发。 显性触发: 基于设备的出口链路, 在每条链 路上配置流量速率阈值, 如果受保证流速超过该阈值, 将超过阈值的报文 设置资源占用告警标志, 出口网关收到标有资源占用告警标志的报文后, 向相关的入口网关发送占用告警消息。 隐性触发: 出口网关检测边到边的 每条路径拥塞等级评估值, 当拥塞等级评估值在预先设定的触发范围时, 出口网关向相应的入口网关发送资源占用告警消息, 设定的触发范围可以 由各运营商根据自身系统的运行情况自行指定的。
参照图 4, 本发明实施例的网段内的出口网关的操作流程如下: 步骤 S31 , 根据标志标签判断收到的报文是否为受保证报文, 如果是非 受保证报文则正常转发, 如果是受保证报文, 则执行步骤 S32;
步驟 S32,基于 MPLS路径统计流速和拥塞等级,统计整个入口网关到 出口网关的拥塞等级;
步骤 S33, 判断告警模式为隐性还是显性, 如果为显性模式, 则执行步 骤 S34, 如果为隐性模式, 则执行步骤 S35;
步骤 S34, 判断是否有占用资源告警标志, 若没有则进到步骤 S37转发 报文到下一跳, 如果有资源占用告警标志, 则执行步骤 S36;
步骤 S35 , 告警模式为隐性模式时, 判断拥塞等级是否超过阀值, 如果 未超过, 则进到步骤 S37, 否则在步驟 S36;
步骤 S36, 向入口网关发送资源占用告警消息;
步骤 S37, 转发报文到下一跳。
转发到下一跳前要去掉受保证标志标签和 MPLS标签。
出口网关向入口网关发送资源占用告警消息大致分以下步骤: 步骤 S41 , 确定入口网关:
出口网关保存有受保证报文的信息, 根据标有资源占用标志报文的到 步骤 S42, 测量受保证报文的数据流速:
出口网关测量从该入口网关到出口网关每条路径上的受保证报文的流 的流速和从该入口网关到出口网关的受保证报文的总流速;
步骤 S43, 向入口网关发送资源占用告警:
基于设备的出口链路, 在每条链路上配置流量速率阔值, 如果受保证 流超过该阈值, 将超过阔值的报文设置资源占用告警标志, 出口网关收到 标有资源占用告警标志的报文后, 向相关的入口网关发送占用告警消息。 出口网关检测边到边的每条路径拥塞等级评估值, 当拥塞等级评估值在预 先设定的触发范围时, 出口网关向相应的入口网关发送资源占用告警消息。 出口网关向入口网关发送资源占用告警消息, 消息包括入口网关到出口网 关每条路径上的可接受的速率和拥塞等级评估值以及边到边的总速率和总 拥塞等级评估值。
入口网关收到资源占用告警后, 测量资源占用告警对应路径的流量, 也可增加对同一边到边其他路径的测量。 当入口网关测得的某路径上的入 口流速大于出口网关测得的流速时, 则说明在边到边的路径上产生丟包现 象, 入口网关需要减少流量。 可以有两种方法, 其一如果同一边到边的其 他路径上的拥塞等级评估值较低, 则将该路径上的一部分流分流到其他路 径上, 其二如果这个边到边的所有路径上没有剩余资源可供利用, 则根据 业务策略中止部分业务。
参见图 5, 本发明实施例中的资源准入控制流程包括如下步骤:
551 ,入口网关收到 RS VP PATH消息,入口网关确定出口网关,将 RS VP PATH消息透明地穿过网段, 转发给出口网关。 并保存该资源请求消息中的 相关信息。
552, 出口网关在 RS VP PATH报文中加入自己的 IP地址, 然后转发给 下一跳;
553 , 出口网关收到来自下游的 RSVP RESV消息时, 将 RSVP预留消 息发送给入口网关, 另外携带入口网关到出口网关每条路径的拥塞等级及 流速, 以及入口网关到出口网关总的拥塞等级及流速;
554, 入口网关根据 RSVP预留信息, 路径上每条路径的流速和拥塞等 级等信息, 确定该资源预留的准入控制以及报文经过哪一条路径。 并在入 口网关保存该受保证报文的相关信息。
图 6所示为本发明实施例中在 MPLS路径上转发的报文的标签栈之前 加入的标志标签的结构, 共 32比特。 图 7所示为本发明实施例中网段内标 志标签从入口路由器到出口路由器的转发过程。
基于端到端的流聚合确定一条 MPLS转发路径; 对于要在上述 MPLS 路径上传送的 ^艮文, 在其标签栈之前加入一个 32比特的标志标签, 该标志 标签的前 20比特用于标签值,取值范围为 MPLS协议的标签预留值 0 ~ 15 , 其剩余的 12比特用于标识报文在路径上经历的转发事件; 由 MPLS路径的 入口标签边缘路由器完成基于流的准入、 分类和策略处理, 封装对应的 MPLS转发标签栈, 并在标签栈之前插入标志标签; 当 艮文经历拥塞、数据 流速率超限等转发事件时, MPLS路径上的标签交换路由器在标志标签上进 行相应的标志; 路径的出口标签边缘路由器, 根据每个报文标识标志的转 发状态, 监测这个路径的状态, 所监测的路径状态可以触发或用作入口标 签边缘路由器进行准入和资源管理的参数。
监测路径的状态包括将一段时间内收到有拥塞标志的报文的比特数除 以这段时间内收到的总的报文的比特数得出路径的拥塞等级。
参见图 8,基于拥塞预警的端到端的质量保证系统主要包括入口网关部 分, 中间路由器部分和出口网关部分。
具体来说, 入口网关部分 100包括: 准入控制装置 101和报文分类判 断装置 102及转发控制装置 103。准入控制装置 101参与前述的结合图 5的 准入控制过程。 报文分类判断装置 102 判断报文是受保证报文还是非受保 证报文, 若为受保证报文, 则正常转发, 否则转发控制装置 103根据设定 的策略进行操作, 比如限速、 标志等, 然后, 先在艮文前面添加 MPLS标 签栈指定该报文的转发路径, 再在标签栈之前添加一个标志标签(该标志 标签标明该报文为受保证报文, 供中间路由器使用) 并进行转发。
中间路由器部分 200·包括报文分类判断装置 201和转发控制装置 202 及拥塞监测装置 203。报文分类判断装置 201判断报文是受保证报文还是非 受保证报文, 若为非受保证报文则正常转发, 否则转发控制装置 202 经过 MPLS标签表明的路径将报文转发到出口网关;拥塞监测装置 203进行基于 出口链路的对报文在出口队列的情况进行监测, 如果用于监测的漏桶字节 数超过阀值, 则按比例给报文设置拥塞标志。 如果是显性告警模式, 中间 路由器部分还包括占用资源告警装置 204,用于将超过阀值的报文设置占用 资源告警标志。
出口网关部分 300包括^ =艮文分类判断装置 301 ,才艮据标志标签判断收到 的报文是否为受保证报文, 若为非受保证报文则正常转发, 否则流速测量 及拥塞等级测量装置 302基于 MPLS路径统计流速及统计拥塞等级, 统计 整个入口网关到出口网关的拥塞等级。 然后占用资源告警生成装置 303 根 据占用资源告警模式的不同分别根据不同的告警模式向入口网关发送占用 资源告警消息或直接转发给下一跳。 具体操作情况如前述结合附图 4 的说 明。
通过上述的说明可以看出, 利用本发明提出的入口网关到出口网关的 资源预留机制, 运营商可以实现动态的根据实时业务的资源分配, 充分利 用了有限的网络资源, 同时也体现了包交换网络资源共享的特性。
应该理解, 上述本发明的实施例公开只是为说明的目的, 本发明要求 的保护范围在所附的权利要求书中予以限定。 在本发明的思想启发下做出 的各种变动和修改都包含在本发明的范围内。

Claims

权 利 要 求
1、 一种实现端到端的质量保证方法, 其特征在于, 包括:
在受保证报文中承载所经历的转发事件信息并设置相应的标识; 统计所述转发事件信息并根据所统计的信息对所述受保证报文进行资 源预留和 /或准入控制。
2、 居权利要求 1所述的质量保证方法, 其特征在于, 通过下述步骤 进行承载所经历的转发事件信息并设置相应的标识:
在基于端到端的流聚合的 MPLS转发路径内, 入口标签边缘路由器根 据资源预留策略确定受保证报文的转发路径, 并为所述受保证报文添加 MPLS标签栈和标志标签;
MPLS 路径上的标签交换路由器在所述受保证报文的标志标签上标志 经历的转发事件信息。
3、 才艮据权利要求 2所述的质量保证方法, 其特征在于, 还包括: 在确定所述受保证报文转发路径之前 , 判断进入报文是否为受保证报 文。
4、 根据权利要求 1至 3中任一项所述的质量保证方法, 其特征在于, 所述统计所述转发事件信息并根据所统计的信息进行资源预留和 /或准入 控制包括如下步驟:
根据每个受保证报文标识所标志的转发状态,获取并监测转发路径的状 态, 并基于所监测的路径状态进行资源预留和 /或准入控制。
5、 根据权利要求 2所述的质量保证方法, 其特征在于, 所述标志标签 位于 MPLS标签栈之前, 所述标志标签包括标签值和事件位, 所述事件位 用于对报文在路径上经历的转发事件进行标识。
6、 根据权利要求 1或 2所述的质量保证方法, 其特征在于, 所述报文 经历的转发事件包括拥塞或 /和所述受保证报文的数据流速率达到或超过预 定阔值。
7、 根据权利要求 4所述的盾量保证方法, 其特征在于, 在所述出口标 签交换路由器监测转发路径的状态的步骤之前, 还包括如下步骤:
当所述 MPLS路径在每一跳均使用不同于其他 MPLS路径的标签时, 出口标签交换路由器根据 MPLS标签确定报文经历的路径。
8、 根据权利要求 4所述的质量保证方法, 其特征在于, 所述获取并监 测转发路径的状态的步驟包括计算路径的拥塞等级, 所述路径的拥塞等级 等于路径上一段时间内收到有拥塞标志的报文的比特数除以这段时间内收 到的受保证 ·ί艮文的比特数。
9、 根据权利要求 4所述的质量保证方法, 其特征在于, 所述基于所监 测的路径状态进行资源预留和 /或准入控制过程, 包括:
检测端到端的每条 MPLS路径的拥塞等级评估值以及端到端的拥塞等 级评估值, 当拥塞等级评估值在预先设定的触发范围时, 触发相应的路由 装置的告警操作。
10、根据权利要求 9所述的质量保证方法, 其特征在于, 所述触发相应 的路由装置的告警操作过程包括:
在 MPLS转发路径的各链路上配置受保证报文的数据流速率阈值, 当 受保证报文的数据流速率超过阈值时, 网段内的路由器向出口网关发送占 用告警信号, 出口网关向相关的入口网关发送占用告警信号。
1 根据权利要求 6所述的质量保证方法, 其特征在于, 还包括: 在所述受保证报文的转发路径中的路由器上设置漏桶,所述漏桶用于监 测设备出口链路的拥塞状态;
根据漏桶字节数为所述受保证报文设置拥塞标志。
12、 4艮据权利要求 11所述的质量保证方法, 其特征在于, 所述监测设 备出口链路的拥塞状态, 为受保证报文设置拥塞标志的步骤包括:
当漏桶字节数低于预定的低阔值时, 受保证报文正常转发;
当漏桶字节数介于预定的低阐值和高阔值之间时,为相应的部分受保证 报文设置拥塞标志;
当漏桶字节数高于预定的高阔值时, 为全部受保证报文设置拥塞标志。
13、根据权利要求 4所述的质量保证方法, 其特征在于, 所述获取并统 计转发路径的状态包括对每条 MPLS路径的受保证报文转发状态进行统计 的步據, 统计 MPLS路径的统计参数和入口网关到出口网关总的统计参数 值, 其中 MPLS路径的统计参数包括 MPLS路径的拥塞等级, 入口网关到 出口网关总的统计参数值包括入口网关到出口网关总的拥塞等级, 所述拥塞等级等于路径上一段时间内收到有拥塞标志的受保证报文的 比特数除以这段时间内收到的受保证报文的比特数。
14、 根据权利要求 13所述的质量保证方法, 其特征在于, 所述 MPLS 路径的统计参数还包括 MPLS路径的数据流速率, 入口网关到出口网关的 统计参数值还包括入口网关到出口网关的总数据流速率。
15、根据权利要求 4所述的质量保证方法, 其特征在于, 所述准入控制 包括资源占用, 所述资源占用的过程包括:
当同一边到边的某条路径的统计参数值大于其他路径的统计参数值时, 将该路径上的一部分业务流分流到其他路径上; 或
' 当同一边到边的所有路径上都没有剩余资源时,根据业务策略中止部分 已准入的业务流。
16、 根据权利要求 2所述的质量保证方法, 其特征在于, 还包括: 根据所述标志标签识别受保证报文, 对每条 MPLS路径的受保证报文 转发状态进行统计, 并根据统计值产生资源占用告警并通知入口网关, 以 便入口网关进行资源管理。
17、 根据权利要求 16所述的质量保证方法, 其特征在于, 通过下述步 骤产生资源占用告警并通知入口网关:
设置隐性触发告警模式,具体包括: 出口网关检测端到端的每条 MPLS 路径的拥塞等级评估值以及总的端到端的拥塞等级评估值, 当拥塞等级评 估值在预先设定的触发范围时, 出口网关向相应的入口网关发送资源占用 告警消息;'或
设置显性触发告警模式,具体包括:在网段内中间路由器的出口链路上 配置受保证报文的流速阈值, 当受保证流超过阈值时, 对相应的报文设置 资源占用告警标志。
18、 一种实现端到端的质量保证系统, 其特征在于, 该系统包括: 入口网关,用于判断进入报文是否为受保证报文时,并在该进入报文为 受保证报文时 >据资源预留策略确定转发路径, 并为所述受保证报文添加 MPLS标签栈和标志标签; 中间路由器,用于将入口网关部分转发来的受保证报文传输到出口网关 部分, 并对所述受保证报文所经历的转发事件进行标记和统计; 和
出口网关,用于才艮据所述标志标签识別受保证报文,去掉所述标志标签 和 MPLS标签, 转发到下一跳;
所述出口网关基于入口网关到出口网关的 MPLS路径的统计参数以及 总的统计参数值产生资源占用信息, 并将所述资源信息发送到入口网关, 以使入口网关进行资源管理, 所述资源管理包括资源预留和 /或准入控制。
19、 根据权利要求 18所述的质量保证系统, 其特征在于, 所述中间路 由器上设置有漏桶, 所述漏桶用于监测设备出口链路的拥塞状态, 所述中 间路由器根据漏桶字节数为受保证报文设置拥塞标志, 当漏桶字节数高于 预定阁值时, 为相应的受保证报文设置拥塞标志。
20、 根据权利要求 18所述的质量保证系统, 其特征在于, 所述入口网 关包括:
准入控制单元, 根据 RSVP预留信息、 各相关路径的流速和拥塞等级 信息, 确定报文可否准入以及报文路径;
报文类型判断单元, 判断报文是否为受保证^ =艮文,
转发控制单元, 根据设定的策略指定报文的转发路径, 并为所述报文 类型判断单元确定的受保证报文设置标志标签进行转发。
21、 根据权利要求 18所述的质量保证系统, 其特征在于, 所述中间路 由器包括报文分类判断单元、 转发控制单元及拥塞监测单元,
报文分类判断单元, 判断报文是否为受保证^ =艮文, 若为受保证报文则 正常转发, 转发控制单元按照 MPLS标签表明的路径将报文转发到出口网 关;
拥塞监测单元, 在出口链路对报文进行监测, 如果用于监测的漏桶字 节数超过阀值, 则按比例给报文设置拥塞标志。
22、 根据权利要求 18所述的质量保证系统, 其特征在于,
出口网关包括报文分类判断单元和流速测量及拥塞等级测量单元; 报文分类判断单元, 根据标志标签判断收到的报文是否为受保证报文, 若为受保证报文则正常转发, 流速测量及拥塞等级测量单元基于 MPLS路 径统计流速及统计拥塞等级, 统计整个入口网关到出口网关的拥塞等级。 23、 根据权利要求 18所述的质量保证系统, 其特征在于,
如果是显性告警模式, 中间路由器部分还包括占用资源告警单元, 用 于将超过阀值的报文设置占用资源告警标志;
所述出口网关还包括占用资源告警生成单元,根据不同的占用资源告警 模式分別向入口网关发送占用资源告警消息或直接转发给下一跳。
PCT/CN2007/001187 2006-07-20 2007-04-12 Procédé et système de garantie de qualité de bout en bout WO2008011778A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CNA2006100617242A CN101110753A (zh) 2006-07-20 2006-07-20 基于拥塞预警的端到端的质量保证方法及系统
CN200610061724.2 2006-07-20

Publications (1)

Publication Number Publication Date
WO2008011778A1 true WO2008011778A1 (fr) 2008-01-31

Family

ID=38981128

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2007/001187 WO2008011778A1 (fr) 2006-07-20 2007-04-12 Procédé et système de garantie de qualité de bout en bout

Country Status (2)

Country Link
CN (1) CN101110753A (zh)
WO (1) WO2008011778A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113872872A (zh) * 2021-09-29 2021-12-31 新华三信息安全技术有限公司 一种报文转发方法、装置及框式网络设备

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013082789A1 (zh) * 2011-12-08 2013-06-13 华为技术有限公司 一种拥塞控制方法及装置
CN105391646A (zh) * 2015-10-19 2016-03-09 上海斐讯数据通信技术有限公司 一种链路层设备预警处理的方法和装置
CN108092866B (zh) * 2016-11-23 2021-07-16 华为技术有限公司 一种隧道切换的方法、设备和系统
CN108235804B (zh) * 2017-12-27 2021-12-31 达闼机器人有限公司 一种网络限速方法、装置和服务器
CN112825511A (zh) * 2019-11-20 2021-05-21 深圳市中兴微电子技术有限公司 一种交换网络拥塞管理方法、装置、设备和存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1392686A (zh) * 2002-07-11 2003-01-22 北京邮电大学 一种自动交换光网络可恢复路径选择建立方法
JP2003244238A (ja) * 2002-02-15 2003-08-29 Kddi Corp トラヒック監視装置及びその方法、コンピュータプログラム
US20040081098A1 (en) * 2002-09-11 2004-04-29 Nec Corporation Resource reservation protocol substitute reply router, resource reservation protocol substitute reply system and resource reservation protocol substitute reply method used in the same
CN1633106A (zh) * 2004-12-16 2005-06-29 中国科学院计算技术研究所 一种具有回溯能力的面向发送方的资源预留实现方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003244238A (ja) * 2002-02-15 2003-08-29 Kddi Corp トラヒック監視装置及びその方法、コンピュータプログラム
CN1392686A (zh) * 2002-07-11 2003-01-22 北京邮电大学 一种自动交换光网络可恢复路径选择建立方法
US20040081098A1 (en) * 2002-09-11 2004-04-29 Nec Corporation Resource reservation protocol substitute reply router, resource reservation protocol substitute reply system and resource reservation protocol substitute reply method used in the same
CN1633106A (zh) * 2004-12-16 2005-06-29 中国科学院计算技术研究所 一种具有回溯能力的面向发送方的资源预留实现方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113872872A (zh) * 2021-09-29 2021-12-31 新华三信息安全技术有限公司 一种报文转发方法、装置及框式网络设备
CN113872872B (zh) * 2021-09-29 2023-11-24 新华三信息安全技术有限公司 一种报文转发方法、装置及框式网络设备

Also Published As

Publication number Publication date
CN101110753A (zh) 2008-01-23

Similar Documents

Publication Publication Date Title
US8724462B2 (en) Congestion handling in a packet switched network domain
US6646988B1 (en) System, device, and method for allocating excess bandwidth in a differentiated services communication network
Oueslati et al. A new direction for quality of service: Flow-aware networking
EP2090035B1 (en) Congestion control in stateless domains
Menth et al. A survey of PCN-based admission control and flow termination
AU2007353984B2 (en) Priority flow handling in stateless domains
US10355974B2 (en) Admission control in a packet network
US20110026399A1 (en) Admission control and routing in a packet network
WO2008011778A1 (fr) Procédé et système de garantie de qualité de bout en bout
Andrikopoulos et al. A fair traffic conditioner for the assured service in a differentiated services internet
CN101350764B (zh) 一种网络流量控制方法
Westberg et al. Resource management in diffserv (RMD) framework
Ahmed et al. A measurement-based approach for dynamic QoS adaptation in DiffServ networks
US20080232360A1 (en) Session Admission Control Method and Apparatus
Ramanathan et al. Active resource management for the differentiated services environment
Gerla et al. Resource allocation and admission control styles in QoS DiffServ networks
Kumar et al. Using edge-to-edge feedback control to make assured service more assured in DiffServ networks
EP2134030B1 (en) System and method for dynamic management of QoS real-time multimedia traffic in packet based Mesh networks
Hoang et al. Fair intelligent congestion control over diffserv: A resource discovery and control scheme for diffserv
Wu et al. Direct congestion control scheme (DCCS) for differentiated services IP networks
Charny et al. A Survey of PCN-Based Admission Control and Flow Termination
Menth Efficiency of PCN-based network admission control with flow termination
JP2004193975A (ja) パケット廃棄方法およびパケット交換装置
Chang et al. Adaptive marking threshold for improving bandwidth assurance in a differentiated services network
Jourjon et al. Implementation and performance analysis of a QoS-aware TFRC mechanism

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 07720760

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

NENP Non-entry into the national phase

Ref country code: RU

122 Ep: pct application non-entry in european phase

Ref document number: 07720760

Country of ref document: EP

Kind code of ref document: A1