WO2010020156A1 - 一种混合业务流的缓存处理方法、存储转发方法及装置 - Google Patents

一种混合业务流的缓存处理方法、存储转发方法及装置 Download PDF

Info

Publication number
WO2010020156A1
WO2010020156A1 PCT/CN2009/073235 CN2009073235W WO2010020156A1 WO 2010020156 A1 WO2010020156 A1 WO 2010020156A1 CN 2009073235 W CN2009073235 W CN 2009073235W WO 2010020156 A1 WO2010020156 A1 WO 2010020156A1
Authority
WO
WIPO (PCT)
Prior art keywords
data packet
control information
port
packet
type
Prior art date
Application number
PCT/CN2009/073235
Other languages
English (en)
French (fr)
Inventor
江帆
Original Assignee
中兴通讯股份有限公司
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 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Priority to US13/059,783 priority Critical patent/US8693472B2/en
Priority to EP09807855.3A priority patent/EP2320609B1/en
Publication of WO2010020156A1 publication Critical patent/WO2010020156A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/10Flow control; Congestion control
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/90Buffering arrangements
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/90Buffering arrangements
    • H04L49/9042Separate storage for different parts of the packet, e.g. header and payload
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/90Buffering arrangements
    • H04L49/9047Buffering arrangements including multiple buffers, e.g. buffer pools

Definitions

  • the present invention relates to the field of Ethernet data service processing, and in particular, to a cache processing method, a storage and forwarding method, and a device for a hybrid service flow.
  • Data processing for multi-port input mixed traffic is a complex and cumbersome task.
  • the same processing method as the single service flow is still used, that is, the input data stream is applied and stored in the order of the input data, and the data of the enqueue is completed.
  • the package is assigned to the corresponding queue, waiting for the output of the scheduling request.
  • Each queue has different output attributes, and according to these attributes, scheduling is performed according to a certain scheduling policy. Therefore, the data packet is read out from the cache, and the data packet is output according to an agreed or standard-compliant format to complete the process of storing and forwarding the data packet.
  • Such a design is capable for a traditional, simple business flow.
  • the present invention provides a cache processing method, a storage and forwarding method, and a device for a mixed service flow, which can simplify the design complexity and reduce the test difficulty in the processing of the mixed service flow.
  • the present invention uses the following technical solutions:
  • a cache processing method for a mixed service flow where the hybrid service flow includes at least one data packet, and the cache processing method includes the following steps:
  • At least one port buffer is configured for each input port that receives the input of the hybrid service flow, and at least one type of buffer is configured for each service type of the hybrid service flow;
  • A2 storing control information of the data packet into a port buffer corresponding to an input port of the data packet
  • A3. Obtain control information of the data packet of the port buffer area, identify, by the control information of the obtained data packet, a service type of the data packet, and store control information of the obtained data packet into the data packet.
  • the type cache corresponding to the type of business.
  • two port buffer areas are respectively configured for each input port that receives the input of the mixed service flow
  • the control information of the data packet includes data packet description information and data packet storage description information, where The packet description information and the packet storage description information are respectively stored in two port buffers corresponding to the input port of the data packet.
  • the method further includes, between the steps A2 and A3, determining, according to the packet control information, whether the data packet is a discarded data packet, and if yes, discarding the data packet.
  • the present invention also discloses a method for storing and forwarding a mixed service flow, where the mixed service flow includes at least one data packet, and the storage and forwarding method includes a data packet buffer process and a data packet scheduling output, where the data packet buffer processing is performed.
  • the steps include:
  • At least one port buffer is configured for each input port that receives the input of the hybrid service flow, and at least one type of buffer is configured for each service type of the hybrid service flow;
  • the system information identifies the service type of the data packet, and stores the acquired control information of the data packet in a type buffer corresponding to the service type of the data packet.
  • steps of the packet scheduling output include:
  • C2 Perform scheduling output on the data packets in each queue according to a predetermined arbitration scheduling policy.
  • the predetermined arbitration scheduling policy is an RR polling scheduling.
  • two port buffer areas are respectively configured for each input port that receives the input of the mixed service flow, and the control information of the data packet includes data packet description information and data packet storage description information, where The packet description information and the packet storage description information are respectively stored in two port buffers corresponding to the input port of the data packet.
  • the step B2 and the B3 further include: determining, according to the data packet control information, whether the data packet is a discarded data packet, and if yes, performing a discarding process on the data packet.
  • the present invention also discloses a store-and-forward device for a mixed service flow, where the hybrid service flow includes at least one data packet, and the store-and-forward device includes a data packet control information cache module, an enqueue pre-processing module, and an enqueue mapping module. among them,
  • the data packet control information cache module is configured to configure at least one port buffer area for each input port that receives the input of the mixed service flow, and store control information of the data packet into an input port of the data packet.
  • Port buffer area For each input port that receives the input of the mixed service flow, and store control information of the data packet into an input port of the data packet.
  • the enqueue pre-processing module is configured to acquire control information of the data packet by the port buffer area, identify a service type of the data packet according to the acquired control information of the data packet, and control information of the acquired data packet Storing a type buffer corresponding to the service type of the data packet; and the enqueue mapping module is configured to obtain control information of the data packet by the type buffer area, and map the data packet to each queue according to a predetermined manner. The data packets in each queue are scheduled and output. Further, the apparatus further includes an enqueue statistics module, and the enqueue statistics module is configured to count the number of data packets of different service types discarded and received.
  • the invention caches the data packet of the mixed service flow into the corresponding buffer area by using the port + type and sequentially waits for the corresponding processing flow.
  • This allows each type of data packet from each port to enjoy an independent processing flow. Therefore, the control flow of different ports and different types of data packets can be processed in parallel, which improves the overall performance index and facilitates control of different types of data.
  • FIG. 1 is a schematic block diagram of a store-and-forward device according to an embodiment of the present invention
  • FIG. 2 is a flowchart of a storage and forwarding method according to an embodiment of the present invention
  • FIG. 3 is a flowchart of a cache processing according to an embodiment of the present invention.
  • the input data packet needs to be subjected to packet header analysis, thereby performing classification processing of the data packet, and obtaining a unicast stream, a multicast stream, a broadcast stream, a flooding stream or other types.
  • data flow The traffic policing of the policy is required for each type of data flow. Only when the traffic policing requirements are met can the packet be processed.
  • Packet information is mainly divided into two aspects: one is the control flow, which mainly includes the control information of the data packet; the other is the data flow, which mainly refers to the content part of the data packet.
  • a store-and-forward device mainly includes an input packet control information cache module, a CPU interface module, an enqueue pre-processing module, and an enqueue mapping module. And the enrollment statistics module.
  • the packet control information is composed of two parts: the packet description information and the packet storage description information.
  • control information the control flow information of the data packet is referred to as control information as a whole, and the information about the data packet attribute and the service type in the control information is referred to as data packet description information, and information about the storage aspect of the data packet in the control information is to be controlled. It is called a packet storage description information.
  • the packet header data is analyzed by the packet header to obtain the related description information of the service type and the packet attribute to which the data packet belongs, and is sent as an input to the input packet control information buffer module; on the other hand, the data content of the data packet needs to be correspondingly Cache and management, so the relevant storage description information of the data packet is obtained, and is sent to the data packet control information cache module as another input.
  • the packet control information cache module manages at least one port buffer area, and buffers the control information of the data packet into the corresponding port buffer area according to the port source of the data packet, and one port corresponds to one port buffer area, and preferably, One port corresponds to two port buffers, one for buffering packet description information in the control information, and one for buffering the packet storage description information in the control information.
  • the CPU interface module mainly accepts control information and configuration data from the CPU control module.
  • the main configuration data includes the service type to which the packet belongs, such as the member lookup table for packets such as multicast, broadcast, and flooding, the queue mapping mode and the queue mapping table, and the statistical results read out by the enqueue statistics module.
  • the enqueue pre-processing module is a key processing part of the apparatus of the present invention. It manages to have at least one type of cache area for buffering packet control information of the corresponding service type.
  • the main functions of the pre-processing part of the enqueue are: Identify the service type to which the data packet belongs, and store it in the buffer of the corresponding service type. At the same time, for all the members of the buffers waiting for the queue of different service types (ie, the control information of each packet), use a certain arbitration and scheduling policy to select the members in the buffer of a certain service type to send the enqueue mapping.
  • the module performs enqueue mapping processing.
  • the process of the pre-processing part of the enqueue can be referred to Figure 2, and the specific process will be described later.
  • the enqueue mapping module whose main function is based on a certain queue mapping principle (for example, a team that can obtain the packet according to the 802.1P, the packet input port number, and the packet output port number. Column number, etc.), all packets in the queue of different service types are mapped to the corresponding queues, so that the data packets of the same queue are stored and managed in the manner of queues; meanwhile, all data flow services They are also scheduled out in the respective queues in the order they are entered.
  • the mapping mode of the queue can be obtained by searching the configuration table in the foregoing CPU interface module to obtain the corresponding queue ID.
  • the data of the data packet itself that is, the content of the data packet is stored in the external buffer space, and then only the start of the data packet in the external cache space needs to be recorded. At the end position, the entire packet content can be located.
  • the description information of the data packet and the location information of the data packet in the cache space ie, the foregoing storage description information
  • the data packet needs to be sent out it only needs to extract the content of the entire data packet according to the storage location information, and perform targeted modification according to the description information of the data packet, that is, the last data sent. Therefore, in this paper, after entering the input packet control information buffer module, the referred data packet is the control information of the data packet (packet description information + storage description information), not the content of the data packet.
  • the enqueue statistics module counts the number of packets discarded and received by different service types; and, the discarded packets need to identify different discard types. In addition, for the dropped packets generated in the pre-processing part of the enqueue, the enqueue statistics module also performs relevant statistics. The role of the enqueue statistics module is to enhance the monitoring of data packets of different service types and improve the testability of modules under mixed service flows.
  • the main process includes: after storing and managing a data packet of a mixed service flow, sending the storage description information of the data packet to the data packet control information cache module. After performing packet header analysis on the data packet, the description information of the data packet is obtained, and then sent to the data packet control information cache module.
  • the control information of the data packet is read from the corresponding buffer area of the data packet control information according to the carried port information. If the data is a dropped data packet, the data flow control part of the data packet is notified, and the number of enqueues of the data packet is set to 0, indicating that the data packet is a discarded data packet, and the related information of the data packet is sent back.
  • the packet data flow control part is based on the drop indication information. The data storage space occupied by the discarded packet is completed, and the released data storage space will be allocated again.
  • the main function of the data flow control part is the allocation and recovery of the storage space.
  • the data content of the data packet needs to be stored in the external cache space, and the location information in the external cache space is recorded; when the data content of the data packet After being read, the storage space occupied by the packet needs to be released for later distribution to store the data content of the new packet.
  • the data flow control portion and the apparatus of the present invention are two adjacent units that are processed in close proximity.
  • the type to which the packet belongs is identified after receiving the packet control information.
  • the packet is written to the buffer of the corresponding service type.
  • Each service type buffer area is only used for temporary storage purposes, and then the corresponding service type of the packet control information is processed, and the processed control information is stored in the next-level data packet enqueue buffer area corresponding to each type.
  • the type 1 branch under the service type decision box directly reaches the type 1 enqueue buffer after the judgment buffer is full, which is slightly different from the descriptions of the types 2, 3, and 4 branches. This is mainly to reflect that the processes handled by different types of business flows may be different.
  • each enqueue buffer must have active information. If there is no active information, it will directly skip the currently scheduled enqueue buffer and enter the next enqueue buffer for scheduling.
  • unicast packets need to write control information of the packets to the enqueue buffer when entering the unicast enqueue buffer.
  • Unicast data stored in the unicast enqueue buffer after queue mapping and other functions are completed The packet is output according to the agreed standard packet descriptor format and according to a certain scheduling policy.
  • Common data service types include unicast, multicast, broadcast, and flooding services. The following four common port flows are taken as an example. The present invention will be further described in detail with reference to FIG.
  • the packet header data is analyzed by the packet header, and the packet description information of the packet type and the packet attribute is obtained, and is stored according to the different packets from the port.
  • the packet description information from port 1 is stored in port 1 buffer area 1
  • the packet description information from port 2 is stored in port 2 buffer area 2
  • the packet description information from port 3 is stored in the port.
  • the packet description information from port 4 is stored in port 4 buffer area 4.
  • the storage description information of the data packet from port 1 is stored in the buffer area 5 of the port 1
  • the storage description information of the data packet from the port 2 is stored in the buffer area 6 of the port 2
  • the storage description information of the data packet from the port 3 is stored in the port 3.
  • Cache area 7 the storage description information of the data packet from port 4 is stored in port 4 buffer area 8.
  • buffer 1 to buffer 8 are implemented in the form of FIFO (first in first out), and each FIFO has a depth of 16. All the buffer areas cannot discard the packet information.
  • the packet description information such as the packet type and packet attributes that have been obtained in the packet header data analysis part must be stored in the buffer area 1 to the buffer area 4 of the corresponding port. .
  • the packet storage description information that has been obtained in the packet buffer processing portion must be stored in the buffer area 5 of the corresponding port to the buffer area 8.
  • the control information of the port 1 can be synthesized, and sent to the type buffer of the corresponding type according to the type of the data packet. Otherwise, if one of the buffers 1 or 5 corresponding to the input port 1 is empty, the synthesis of the packet control information cannot be performed.
  • the principle of packet control information for Port 2, Port 3, and Port 4 is the same as for Port 1.
  • the packet After the packet control information enters the corresponding type buffer, the packet can be arbitrated. And the scheduling, arbitration and scheduling strategies are more flexible, and the simplest and most commonly used RR polling scheduling method is used here. The processing manner of the RR polling scheduling is well known to those skilled in the art, and details are not described herein again.
  • a multicast packet to be processed is taken out from the multicast type buffer, and the member first pointer of the multicast group is extracted from the control information of the multicast packet.
  • the multicast member table is searched for by the extracted multicast group first pointer, and all members of the multicast group and the information to be modified or replaced are obtained.
  • the packet descriptor information after completion of multicast replication is written into the multicast enqueue buffer. After each replication completes a multicast member, you need to check whether all members of the current multicast group are copied. If the member replication has not been completed, continue to copy. When the replication is complete, repeat the previous operation (that is, re-type from the type cache). The multicast packet is taken out to start a new copy process until all members have copied.
  • the corresponding arbitration and scheduling processing strategies can be flexibly selected.
  • the example described above is an operation example of four ports, and it does not only mean that only four ports of data streams can be processed, and thus the processing of the input data stream extended to any number of ports can be performed.
  • the four different service types of unicast packets, multicast packets, broadcast packets, and flood packets are listed above, it does not only mean that only four types of packet data streams can be processed, and thus can be extended. Processing to other types of packets.
  • the invention discloses a method and a device for implementing a mixed service flow buffer processing, storing and forwarding, which can process different types of data streams in parallel, and improve the performance of processing data streams,
  • the ratio of different types of data streams to the overall bandwidth can be allocated.
  • the design process is simple and clear.
  • it is convenient to detect the real-time situation of different data streams in the business process, enhance the testability, and greatly reduce the difficulty of the test.
  • Different problems involving different types of data streams and the requirements of the proportion of different data streams in the overall bandwidth can be better solved.
  • the invention has good scalability.
  • the data stream services that need to be processed are becoming more and more abundant, and the bandwidth is getting higher and higher, which can be easily extended on the basis of the present invention, expanding to more input service flow ports, and more
  • different scheduling strategies can be flexibly used according to the types of different services and bandwidth requirements.
  • the present invention can process different types of data streams in parallel, and can improve the performance of processing data streams, and can also allocate the proportion of different types of data streams to the overall bandwidth. Different problems involving different types of data streams and the requirements of the proportion of different data streams in the overall bandwidth can be better solved.
  • the invention has good scalability. It is extended to the design of more input service flow ports and more service types, and it is also possible to flexibly adopt different scheduling strategies according to the types of services and bandwidth requirements.

Description

一种混合业务流的緩存处理方法、 存储转发方法及装置
技术领域
本发明涉及以太网数据业务处理领域, 具体的说涉及一种混合业务流的 緩存处理方法、 存储转发方法及装置。
背景技术
对多端口输入混合业务流的数据处理, 是一项比较复杂和麻烦的工作。 目前, 在对多端口输入的混合业务流进行处理的过程中, 依然釆用和单一业 务流一样的处理方式, 即输入的数据流按照输入的顺序进行数据緩存申请和 存储, 完成入队的数据包被分配到相应的队列当中, 等待输出的调度申请。 每个队列有不同的输出属性, 根据这些属性按照一定的调度策略进行调度处 理。 从而将数据包从緩存中读出, 并按照约定或符合标准的格式将数据包输 出, 以完成数据包的存储转发的处理过程。 这样的设计, 对于传统、 简单的业务流而言, 是能够胜任的。 但是, 伴 随着数据业务的飞速发展, 这样的处理已经显得有些捉襟见肘了, 既不能满 足混合业务处理的性能要求, 也不利于对某一业务的处理和监测。 同时, 由 于混合业务流的业务类型众多, 业务输入的端口数量也呈增加之势, 现有技 术中这种仅按照输入顺序进行数据緩存的方式导致对混合业务流数据包的测 试复杂度增加, 测试的难度比较大。
发明内容
有鉴于上述背景, 本发明提供了一种混合业务流的緩存处理方法、 存储 转发方法及装置, 可以在对混合业务流的处理中, 在简化设计的复杂性同时 降低测试难度。 为了解决上述技术问题, 本发明釆用了如下技术方案:
一种混合业务流的緩存处理方法,所述混合业务流包括至少一个数据包, 所述緩存处理方法包括如下步骤:
A1、为接收所述混合业务流输入的每一输入端口分别配置至少一个端口 緩存区,为所述混合业务流的每一种业务类型分别配置至少一个类型緩存区;
A2、将数据包的控制信息存入到所述数据包的输入端口所对应的端口緩 存区; 以及
A3、 获取所述端口緩存区的数据包的控制信息, 由所获取的数据包的控 制信息识别所述数据包的业务类型, 并将所获取的数据包的控制信息存入到 所述数据包的业务类型所对应的类型緩存区。
进一步地, 所述方法中, 为接收所述混合业务流输入的每一输入端口分 别配置两个端口緩存区, 所述数据包的控制信息包括数据包描述信息和数据 包存储描述信息, 所述数据包描述信息和数据包存储描述信息分别存在所述 数据包的输入端口所对应的两个端口緩存区中。
进一步地, 所述方法在所述步骤 A2和 A3之间还包括: 根据数据包控制 信息判断数据包是否为丟弃数据包, 如是, 对该数据包进行丟弃处理。
本发明还公开了一种混合业务流的存储转发方法, 所述混合业务流包括 至少一个数据包,所述存储转发方法包括数据包緩存处理和数据包调度输出, 其中, 数据包緩存处理的所述步骤包括:
B1、 为接收所述混合业务流输入的每一输入端口分别配置至少一个端口 緩存区,为所述混合业务流的每一种业务类型分别配置至少一个类型緩存区;
B2、 将数据包的控制信息存入到所述数据包的输入端口所对应的端口緩 存区; 以及
B3、 获取所述端口緩存区的数据包的控制信息, 由所获取的数据包的控 制信息识别所述数据包的业务类型, 并将所获取的数据包的控制信息存入所 述数据包的业务类型所对应的类型緩存区。
进一步地, 数据包调度输出的所述步骤包括:
Cl、 提取业务类型緩存区中的数据包控制信息, 根据所述数据包控制信 息获取数据包并将获取的数据包映射到各个队列中; 以及
C2、 按预定的仲裁调度策略, 对所述各队列中的数据包进行调度输出。 进一步地, 所述方法中, 所述预定的仲裁调度策略为 RR轮询调度。 进一步地, 所述方法中, 为接收所述混合业务流输入的每一输入端口分 别配置两个端口緩存区, 所述数据包的控制信息包括数据包描述信息和数据 包存储描述信息, 所述数据包描述信息和数据包存储描述信息分别存在所述 数据包的输入端口所对应的两个端口緩存区中。
进一步地, 所述方法中, 所述步骤 B2和 B3之间还包括: 根据数据包控 制信息判断数据包是否为丟弃数据包, 如是, 对该数据包进行丟弃处理。
本发明同时公开了一种混合业务流的存储转发装置, 所述混合业务流包 括至少一个数据包, 所述存储转发装置包括数据包控制信息緩存模块、 入队 预处理模块和入队映射模块, 其中,
所述数据包控制信息緩存模块设置成为接收所述混合业务流输入的每一 输入端口分别配置至少一个端口緩存区, 并将数据包的控制信息存入到所述 数据包的输入端口所对应的端口緩存区;
所述入队预处理模块设置成由所述端口緩存区获取数据包的控制信息, 根据所获取的数据包的控制信息识别所述数据包的业务类型, 并将所获取的 数据包的控制信息存入到所述数据包的业务类型所对应的类型緩存区; 以及 所述入队映射模块用于由所述类型緩存区获取数据包的控制信息, 按预 定方式将数据包映射到各个队列, 并对各个队列中的数据包进行调度输出。 进一步地, 所述装置还包括入队统计模块, 所述入队统计模块设置成统 计丟弃和接收的不同业务类型的数据包的数目。 本发明依据输入业务流的输入端口和业务类型的不同,釆用端口 +类型的 方式将混合业务流的数据包緩存到对应的緩存区中并依次等待相应的处理流 程。 这使得来自各端口的各类型的数据包享有独立的处理流程, 因此, 不同 端口、 不同类型的数据包的控制流可以并行地处理, 既提高了整体的性能指 标, 又便于控制不同类型的数据包所占用的整体带宽的比例, 同时也具备良 好的扩展性。
附图概述
图 1是本发明具体实施方式的存储转发装置的原理框图;
图 2是本发明具体实施方式的存储转发方法的流程图; 图 3是本发明具体实施方式的緩存处理的流程图。
本发明的较佳实施方式
下面对照附图并结合具体实施方式对本发明做进一步详细说明。 对于包含多个数据包的混合业务流的处理, 需要对输入的数据包进行包 头分析, 从而进行数据包的分类处理, 得到单播流、 组播流、 广播流以及洪 泛流或者其他类型的数据流。 每种数据流都需要进行 Policing的流量监管, 只有满足流量监管的要求才能进行数据包的入队处理。 数据包信息主要分为两个方面: 一个是控制流, 主要包括数据包的控制 信息; 另一个是数据流, 主要是指数据包的内容部分。 在本发明中, 主要涉 及数据包的控制信息部分。 如图 1所示, 本发明具体实施方式的存储转发装置, 主要包括输入数据 包控制信息緩存模块, CPU接口模块, 入队预处理模块, 入队映射模块, 以 及入队统计模块。 数据包控制信息由数据包描述信息和数据包存储描述信息 两部分组成。 在此, 将数据包的控制流信息从整体上称为控制信息, 将控制 信息中有关数据包属性和业务类型的信息称为数据包描述信息, 将控制信息 中有关数据包的存储方面的信息称为数据包存储描述信息。 一方面, 包头数 据经过包头分析,得到数据包所属的业务类型及数据包属性等相关描述信息, 作为输入送给输入数据包控制信息緩存模块; 另一方面, 数据包的数据内容 需要进行相应的緩存和管理, 因此会得到数据包的相关存储描述信息, 作为 另一个输入送给数据包控制信息緩存模块。 数据包控制信息緩存模块管理有 至少一个端口緩存区, 将数据包的控制信息按其数据包的端口来源緩存到相 应的端口緩存区中, 可以一个端口对应一个端口緩存区, 优选的, 还可以一 个端口对应两个端口緩存区, 一个用于緩存控制信息中的数据包描述信息, 一个用于緩存控制信息中的数据包存储描述信息。
CPU接口模块主要是接受来自 CPU控制模块的控制信息及配置数据。 主 要的配置数据包括数据包所属业务类型, 例如组播、 广播和洪泛等数据包的 成员查找表, 队列映射模式及队列映射表, 以及由入队统计模块中读出的统 计结果。
入队预处理模块, 是本发明装置的关键处理部分。 其管理有至少一个类 型緩存区, 用于緩存对应业务类型的数据包控制信息。 入队预处理部分主要 完成的功能有: 对数据包所属的业务类型进行识别, 并存入到对应业务类型 的緩存区中。 同时, 对于不同业务类型的所有等待入队的緩存区的成员 (即 各数据包的控制信息) , 釆用一定的仲裁和调度策略, 选择某一业务类型緩 存区中的成员送给入队映射模块进行入队映射处理。 入队预处理部分的处理 流程可以参照图 2 , 具体过程将在后文描述。
入队映射模块, 其主要功能是根据一定的队列映射原则 (例如可以根据 802. 1P、 数据包输入端口号和数据包输出端口号进行组合获得该数据包的队 列号等等) , 将不同业务类型入队緩存区中所有的数据包映射到相应的队列 中, 以便于相同队列的数据包釆用队列的方式进行存储和管理; 同时, 所有 的数据流业务也在各自队列中按照输入的顺序被调度出去。 队列的映射方式 可以通过查找前述 CPU接口模块中的配置表, 得到相应的队列 ID。 需要理解 的是, 对于数据包处理, 是在数据包到来后, 将数据包本身的数据即数据包 的内容存到外部緩存空间, 之后只需要记录该数据包在外部緩存空间中的起 始和结束位置, 就可以定位到整个数据包内容, 在后续的处理过程中, 管理 和传递的只是这个数据包的描述信息和该数据包在緩存空间中的位置信息 (即前述存储描述信息) , 当这个数据包需要发送出去时, 只需要根据存储 位置信息, 将整个数据包的内容提取出来, 并按照数据包的描述信息做针对 性的修改, 就是最后发送出去的数据了。 因此在本文中, 从进入到输入数据 包控制信息緩存模块后,所指的数据包就是数据包的控制信息(包描述信息 + 存储描述信息) , 而不是指数据包的内容。
入队统计模块, 统计丟弃和接收的不同业务类型的数据包数目; 并且, 被丟弃的数据包需要识别出不同的丟弃类型。 另外, 对于在入队预处理部分 产生的丟弃包, 入队统计模块也进行相关的统计。 入队统计模块的作用在于 加强不同业务类型数据包的监控, 提高混合业务流下模块的可测性。
图 2所示为本发明具体实施方式的存储转发流程图, 其主要过程包括: 在对混合业务流的数据包进行存储管理之后, 将数据包的存储描述信息 送入到数据包控制信息緩存模块, 在对数据包进行包头分析后, 得到数据包 的描述信息后, 同样送入到数据包控制信息緩存模块。
在经过包緩存处理后, 根据携带的端口信息从数据包控制信息相应的緩 存区中读出数据包的控制信息。 如果该数据为丟弃数据包, 则通知数据包数 据流控制部分, 并设置数据包的入队次数为 0, 指出该数据包为丟弃数据包, 同时送回该数据包的相关信息,由数据包数据流控制部分根据丟弃指示信息, 完成被丟弃数据包所占用的数据存储空间的回收, 释放后的数据存储空间将 被再次分配使用。 数据流控制部分的主要功能是存储空间的分配和回收, 如 前所述, 数据包的数据内容需要存放到外部緩存空间, 并记录下在外部緩存 空间中的位置信息; 当数据包的数据内容被读走以后, 该数据包所占用的存 储空间需要被释放, 以便在以后被分配以存储新的数据包的数据内容。 从结 构上来说,该数据流控制部分和本发明所述装置是处理紧邻的两个独立单元。
如果该数据不为丟弃数据包, 则在接收到数据包控制信息后, 对数据包 所属的类型进行识别。 根据数据包的不同业务类型, 数据包写入到相应业务 类型的緩存区中。 各业务类型緩存区只是作为临时存放用途, 然后进行数据 包控制信息的相应业务类型的处理, 将处理完的控制信息存放在各自类型所 对应的下一级数据包入队緩存区中。 需要注意的是, 在图 2中, 在业务类型 判定框下的类型 1分支在判断緩存区满后直接到达类型 1入队緩冲区, 与类 型 2、 3、 4各分支的描述略有不同, 这主要是为了体现不同类型的业务流所 处理的流程可能是不同的。
在数据包写入到类型緩存区时, 如果类型緩存区满, 则直接丟弃该数据 包。 类型緩存区满丟弃的具体操作, 和数据包为丟弃数据包时候的丟弃操作 是完全一样的。
在不同业务类型的入队緩存区中有控制信息存在时, 就需要进行入队緩 存区的信息提取, 进行队列映射及其他功能的处理。 这就需要对不同业务类 型的入队緩存区进行仲裁和调度处理。 考虑到仲裁调度的效率, 每个入队緩 存区都要有激活的信息, 如果没有激活的信息, 将直接跳过当前被调度的入 队緩存区而进入到下一个入队緩存区进行调度。
尽管不像组播、 广播包或洪泛包一样涉及到多个数据转发目的地址, 单 播包在进入单播入队緩存区时, 也需要将数据包的控制信息写入到入队緩存 区中。 在完成队列映射及其他功能后, 存放在单播入队緩存区中的单播数据 包按照约定的标准包描述符格式并依据一定的调度策略进行输出。 常见的数据业务类型有单播、 组播、 广播和洪泛业务, 下面以这四种常 见的业务流从 4个端口输入为例, 对照图 3对本发明作进一步的详细介绍。
如图 3所示, 来自 4个端口的数据流, 一方面包头数据经过包头分析, 得到数据包的类型及数据包属性等相关的数据包描述信息, 根据数据包来自 端口的不同, 分别存放在不同的端口緩存区中, 来自端口 1的数据包描述信 息存放在端口 1緩存区 1 ,来自端口 2的数据包描述信息存放在端口 2緩存区 2 , 来自端口 3的数据包描述信息存放在端口 3緩存区 3 , 来自端口 4的数据 包描述信息存放在端口 4緩存区 4。
另一方面, 来自 4个端口的数据流, 数据包的内容需要进行相应的緩存 和管理, 因此得到数据包的相关存储方面的存储描述信息。 来自端口 1的数 据包的存储描述信息存放在端口 1緩存区 5 , 来自端口 2的数据包的存储描 述信息存放在端口 2緩存区 6 , 来自端口 3的数据包的存储描述信息存放在 端口 3緩存区 7 ,来自端口 4的数据包的存储描述信息存放在端口 4緩存区 8。
这里,緩存区 1到緩存区 8以 FIFO (先进先出)的形式来实现,每个 FIFO 的深度为 16。 所有的緩存区不能进行数据包信息的丟弃, 已经在数据包包头 数据分析部分完成操作获得的包类型及包属性等数据包描述信息必存入到对 应端口的緩存区 1到緩存区 4中。 同理, 已经在数据包緩存处理部分完成操 作获得的数据包存储描述信息必存入到对应端口的緩存区 5到緩存区 8中。
只有在输入端口 1对应的緩存区 1和緩存区 5都不为空时, 才能合成端 口 1的控制信息, 并根据该数据包的类型送往对应类型的类型緩存区。 否则, 如果输入端口 1对应的緩存区 1或 5有一个为空时, 就不能进行数据包控制 信息的合成。 端口 2、 端口 3和端口 4的数据包控制信息的合成原则与端口 1 相同。
在数据包控制信息进入到对应的类型緩存区后, 可以对数据包进行仲裁 及调度处理, 仲裁及调度处理的策略比较灵活多样, 这里釆用的是最简单也 是最常用的 RR轮询调度方式。 RR轮询调度的处理方式为本领域技术人员所 熟知, 在此不再赘述。
以组播包为例, 当组播入队緩存区不满时, 从组播类型緩存区取出一个 待处理的组播包, 从该组播包的控制信息中提取出组播组的成员首指针, 通 过提取的组播组首指针查找组播成员表, 得到该组播组的所有成员及待修改 或替换信息。
根据组播成员表的配置信息, 再和输入的控制信息一起合成数据包新的 控制信息 (即原来的控制信息 + 成员的配置信息 = 成员新的控制信息) 。 完成组播复制后的包描述符信息写入到组播入队緩存区中。 每复制完成一个 组播成员, 需要检测当前组播组所有成员是否复制完成, 在成员复制尚没有 完成的情况下, 继续复制; 当复制完成后, 重复前面的操作 (即重新从类型 緩存区中取出组播包开始新的复制过程, 直到所有的成员复制完成为止) 。
当组播成员入队緩存区满时, 停止当前组播成员的复制, 直到组播成员 入队緩存区不满时继续复制流程的处理。 广播包、 洪泛包和组播包的复制和 处理原则是相同的。
对于入队緩存区的仲裁和调度处理, 根据不同类型的数据包以及对应的 带宽的要求的不同, 可以灵活地选择相应的仲裁和调度处理策略。
上述描述的实例为 4个端口的操作实例, 并不仅仅意味着只能处理 4个 端口的数据流, 可以由此进行扩展到任意多个端口的输入数据流的处理。 同 样的, 虽然上文只列举单播包、 组播包、 广播包和洪泛包 4种不同业务类型, 也并不仅仅意味着只能处理 4个类型的包数据流, 可以由此进行扩展到其他 类型数据包的处理。
本发明公开了一种实现混合业务流緩存处理、 存储转发的方法和装置, 可以并行地处理不同类型的数据流, 在提升了处理数据流的性能的同时, 亦 可对不同的类型数据流占整体带宽的比例进行分配。 设计流程简单、 清晰, 同时, 通过入队统计模块的设置, 可以方便地检测不同数据流在业务处理过 程中的实时情况, 增强了可测性, 极大地降低了测试的难度。 可以较好地解 决涉及不同类型的数据流以及不同的数据流在整体带宽中所占用比例的要求 的不同的问题。
与此同时, 本发明具有很好的扩展性。 随着通讯行业的不断发展, 需要 处理的数据流业务越来越丰富, 带宽越来越高的情况下, 可以很方便在本发 明的基础上进行扩展, 扩展到更多输入业务流端口、 更多业务类型的设计当 中, 而且也可以根据不同业务的类型及带宽的要求, 灵活地釆用不同的调度 策略。
以上内容是结合具体的优选实施方式对本发明所作的进一步详细说明, 不能认定本发明的具体实施只局限于这些说明。 对于本发明所属技术领域的 普通技术人员来说, 在不脱离本发明构思的前提下, 还可以做出若干简单推 演或替换, 都应当视为属于本发明的保护范围。
工业实用性
本发明可以并行地处理不同类型的数据流, 在提升了处理数据流的性能 的同时, 亦可对不同的类型数据流占整体带宽的比例进行分配。 可以较好地 解决涉及不同类型的数据流以及不同的数据流在整体带宽中所占用比例的要 求的不同的问题。 与此同时, 本发明具有很好的扩展性。 扩展到更多输入业 务流端口、 更多业务类型的设计当中, 而且也可以根据不同业务的类型及带 宽的要求, 灵活地釆用不同的调度策略。

Claims

权 利 要 求 书
1. 一种混合业务流的緩存处理方法, 所述混合业务流包括至少一个数据 包, 所述緩存处理方法包括如下步骤:
A1、为接收所述混合业务流输入的每一输入端口分别配置至少一个端口 緩存区,为所述混合业务流的每一种业务类型分别配置至少一个类型緩存区;
A2、将所述数据包的控制信息存入到所述数据包的输入端口所对应的端 口緩存区; 以及
A3、 获取所述端口緩存区的数据包的控制信息, 由所获取的数据包的控 制信息识别所述数据包的业务类型, 并将所获取的数据包的控制信息存入到 所述数据包的业务类型所对应的类型緩存区。
2. 如权利要求 1所述的緩存处理方法, 其中,
为接收所述混合业务流输入的每一输入端口分别配置两个端口緩存区, 所述数据包的控制信息包括数据包描述信息和数据包存储描述信息, 所述数 据包描述信息和数据包存储描述信息分别存在所述数据包的输入端口所对应 的两个端口緩存区中。
3. 如权利要求 2所述的緩存处理方法, 其中, 所述方法在步骤 A2和 A3 之间还包括:根据所述数据包的控制信息判断所述数据包是否为丟弃数据包, 如是, 对所述数据包进行丟弃处理。
4. 一种混合业务流的存储转发方法, 所述混合业务流包括至少一个数据 包, 所述方法包括数据包緩存处理和数据包调度输出, 其中, 数据包緩存处 理的所述步骤包括:
B1、 为接收所述混合业务流输入的每一输入端口分别配置至少一个端口 緩存区,为所述混合业务流的每一种业务类型分别配置至少一个类型緩存区; B2、 将数据包的控制信息存入到所述数据包的输入端口所对应的端口緩 存区; 以及
B3、 获取所述端口緩存区的数据包的控制信息, 由所获取的数据包的控 制信息识别所述数据包的业务类型, 并将所获取的数据包的控制信息存入所 述数据包的业务类型所对应的类型緩存区。
5. 如权利要求 4所述的存储转发方法, 其中, 数据包调度输出的所述 步骤包括:
Cl、 提取业务类型緩存区中的数据包控制信息, 根据所述数据包控制信 息获取数据包并将所获取的数据包映射到各个队列中; 以及
C2、 按预定的仲裁调度策略, 对所述各队列中的数据包进行调度输出。
6. 如权利要求 5所述的存储转发方法, 其中, 所述预定的仲裁调度策略 为 RR轮询调度。
7. 如权利要求 4所述的存储转发方法, 其中, 为接收所述混合业务流输 入的每一输入端口分别配置两个端口緩存区, 所述数据包的控制信息包括数 据包描述信息和数据包存储描述信息, 所述数据包描述信息和数据包存储描 述信息分别存在所述数据包的输入端口所对应的两个端口緩存区中。
8. 如权利要求 7所述的存储转发方法, 其中, 所述方法在所述步骤 B2 和 B3之间还包括:根据所述数据包的控制信息判断所述数据包是否为丟弃数 据包, 如是, 对所述数据包进行丟弃处理。
9. 一种混合业务流的存储转发装置, 所述混合业务流包括至少一个数据 包, 所述存储转发装置包括数据包控制信息緩存模块、 入队预处理模块和入 队映射模块, 其中,
所述数据包控制信息緩存模块设置成为接收所述混合业务流输入的每一 输入端口分别配置至少一个端口緩存区, 并将数据包的控制信息存入到所述 数据包的输入端口所对应的端口緩存区; 所述入队预处理模块设置成由所述端口緩存区获取数据包的控制信息, 根据所获取的数据包的控制信息识别所述数据包的业务类型, 并将所获取的 数据包的控制信息存入到所述数据包的业务类型所对应的类型緩存区; 以及 所述入队映射模块用于由所述类型緩存区获取数据包的控制信息, 按预 定方式将数据包映射到各个队列, 并对各个队列中的数据包进行调度输出。
10. 如权利要求 9所述的存储转发装置, 其中, 所述装置还包括入队统 计模块, 所述入队统计模块设置成统计丟弃和接收的不同业务类型的数据包 的数目。
PCT/CN2009/073235 2008-08-19 2009-08-13 一种混合业务流的缓存处理方法、存储转发方法及装置 WO2010020156A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US13/059,783 US8693472B2 (en) 2008-08-19 2009-08-13 Buffer processing method, a store and forward method and apparatus of hybrid service traffic
EP09807855.3A EP2320609B1 (en) 2008-08-19 2009-08-13 A buffer processing method, a store and forward method and apparatus of hybrid service traffic

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN200810142472.5 2008-08-19
CN2008101424725A CN101656659B (zh) 2008-08-19 2008-08-19 一种混合业务流的缓存处理方法、存储转发方法及装置

Publications (1)

Publication Number Publication Date
WO2010020156A1 true WO2010020156A1 (zh) 2010-02-25

Family

ID=41706856

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2009/073235 WO2010020156A1 (zh) 2008-08-19 2009-08-13 一种混合业务流的缓存处理方法、存储转发方法及装置

Country Status (4)

Country Link
US (1) US8693472B2 (zh)
EP (1) EP2320609B1 (zh)
CN (1) CN101656659B (zh)
WO (1) WO2010020156A1 (zh)

Families Citing this family (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8965999B1 (en) * 2006-04-20 2015-02-24 At&T Intellectual Property I, L.P. Distribution scheme for subscriber-created content, wherein the subscriber-created content is rendered for a recipient device by the service provider network based on a device characteristic and a connection characteristic of the recipient device
CN102025653B (zh) * 2010-06-04 2014-01-08 西本新干线电子商务有限公司 企业服务总线及企业服务总线的消息处理方法
US8805938B2 (en) 2010-06-04 2014-08-12 Xiben New Line Stock Co., Ltd. Enterprise service bus and message processing method thereof
CN102025649A (zh) * 2010-06-04 2011-04-20 西本新干线股份有限公司 企业服务总线的消息处理方法
CN101950261B (zh) * 2010-09-09 2013-12-04 中兴通讯股份有限公司 数据存储与鉴权并行的处理方法和终端
DE112012000487T5 (de) * 2012-04-10 2014-01-23 Fuji Electric Co., Ltd Leistungsumwandlungseinrichtung
CN102970226B (zh) * 2012-11-12 2015-07-22 盛科网络(苏州)有限公司 在asic中进行组播报文转发的方法和装置
US11003459B2 (en) 2013-03-15 2021-05-11 Intel Corporation Method for implementing a line speed interconnect structure
CN105190579B (zh) * 2013-03-15 2018-04-06 英特尔公司 一种用于实现线路速度互连结构的方法
US9753691B2 (en) 2013-03-15 2017-09-05 Intel Corporation Method for a stage optimized high speed adder
KR102070149B1 (ko) * 2013-06-10 2020-01-28 에스케이텔레콤 주식회사 통신 네트워크에서 컨텐츠 전달 방법 및 이를 위한 장치
CN105099932B (zh) * 2015-06-30 2018-06-05 华为技术有限公司 基于直通转发技术的调度方法及装置
CN106850457B (zh) * 2015-12-04 2019-11-15 中国电信股份有限公司 缓存共享方法和装置
CN105635000B (zh) * 2015-12-30 2019-02-01 华为技术有限公司 一种报文存储转发方法和电路及设备
CN106453137B (zh) * 2016-09-20 2019-04-05 北京航空航天大学 一种以缓解HOL Blocking为目标的动态组播入队方法
CN106954103B (zh) * 2017-03-15 2020-04-28 烽火通信科技股份有限公司 一种100g epon系统中下行业务的发送系统及其发送方法
CN107704573A (zh) * 2017-09-30 2018-02-16 山东浪潮通软信息科技有限公司 一种与业务耦合的智能缓存方法
CN109324761A (zh) * 2018-10-09 2019-02-12 郑州云海信息技术有限公司 一种数据缓存方法、装置、设备及存储介质
CN109947562A (zh) * 2019-03-01 2019-06-28 上海七印信息科技有限公司 一种任务分发限流系统及其任务分配方法
CN111198892A (zh) * 2019-12-24 2020-05-26 泰康保险集团股份有限公司 信息处理方法、装置、电子设备及存储介质
US20230198676A1 (en) * 2021-12-17 2023-06-22 Juniper Networks, Inc. Packet drop monitoring in a virtual router

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1826768A (zh) * 2003-08-25 2006-08-30 思科技术公司 利用动态资源分配进行的大规模排队的可扩展方法
CN1859275A (zh) * 2005-07-27 2006-11-08 华为技术有限公司 多端口以太网交换装置及数据传输方法
CN1913486A (zh) * 2005-08-10 2007-02-14 中兴通讯股份有限公司 一种增强协议报文安全的方法和装置
US20070195761A1 (en) * 2006-02-21 2007-08-23 Cisco Technology, Inc. Pipelined packet switching and queuing architecture

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6522628B1 (en) * 1999-03-01 2003-02-18 Cisco Technology, Inc. Method and system for managing transmission resources in a wireless communication network
US7369550B1 (en) * 1999-04-22 2008-05-06 Advanced Micro Devices Method and apparatus for locking a table in a network switch
US6335938B1 (en) * 1999-05-05 2002-01-01 Advanced Micro Devices, Inc. Multiport communication switch having gigaport and expansion ports sharing the same time slot in internal rules checker
US6816488B1 (en) * 1999-05-21 2004-11-09 Advanced Micro Devices, Inc. Apparatus and method for processing data frames in a network switch
US6625146B1 (en) * 1999-05-28 2003-09-23 Advanced Micro Devices, Inc. Method and apparatus for operating a network switch in a CPU-less environment
US6977930B1 (en) * 2000-02-14 2005-12-20 Cisco Technology, Inc. Pipelined packet switching and queuing architecture
US7342942B1 (en) * 2001-02-07 2008-03-11 Cortina Systems, Inc. Multi-service segmentation and reassembly device that maintains only one reassembly context per active output port
US7016352B1 (en) * 2001-03-23 2006-03-21 Advanced Micro Devices, Inc. Address modification within a switching device in a packet-switched network
US20030014483A1 (en) * 2001-04-13 2003-01-16 Stevenson Daniel C. Dynamic networked content distribution
US20050141503A1 (en) * 2001-05-17 2005-06-30 Welfeld Feliks J. Distriuted packet processing system with internal load distributed
US7042842B2 (en) * 2001-06-13 2006-05-09 Computer Network Technology Corporation Fiber channel switch
US7027414B2 (en) * 2001-08-09 2006-04-11 Hughes Network Systems, Llc Method, apparatus, and system for identifying and efficiently treating classes of traffic
US7716314B1 (en) * 2003-12-30 2010-05-11 Dinochip, Inc. Traffic management in digital signal processor
US7760719B2 (en) * 2004-06-30 2010-07-20 Conexant Systems, Inc. Combined pipelined classification and address search method and apparatus for switching environments
CN1327674C (zh) * 2005-02-25 2007-07-18 清华大学 核心路由器上支持访问控制列表功能的双栈兼容路由查找器
US7440405B2 (en) * 2005-03-11 2008-10-21 Reti Corporation Apparatus and method for packet forwarding with quality of service and rate control
US7873048B1 (en) * 2005-12-02 2011-01-18 Marvell International Ltd. Flexible port rate limiting
US7724754B2 (en) * 2006-02-24 2010-05-25 Texas Instruments Incorporated Device, system and/or method for managing packet congestion in a packet switching network
CN100498740C (zh) * 2007-09-11 2009-06-10 腾讯科技(深圳)有限公司 一种数据缓存处理方法、系统及数据缓存装置
CN101222443B (zh) * 2008-01-30 2012-04-25 杭州华三通信技术有限公司 一种处理报文的方法和网络设备
US7979671B2 (en) * 2008-07-28 2011-07-12 CacheIQ, Inc. Dual hash indexing system and methodology

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1826768A (zh) * 2003-08-25 2006-08-30 思科技术公司 利用动态资源分配进行的大规模排队的可扩展方法
CN1859275A (zh) * 2005-07-27 2006-11-08 华为技术有限公司 多端口以太网交换装置及数据传输方法
CN1913486A (zh) * 2005-08-10 2007-02-14 中兴通讯股份有限公司 一种增强协议报文安全的方法和装置
US20070195761A1 (en) * 2006-02-21 2007-08-23 Cisco Technology, Inc. Pipelined packet switching and queuing architecture

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP2320609A4 *

Also Published As

Publication number Publication date
CN101656659B (zh) 2012-05-23
CN101656659A (zh) 2010-02-24
EP2320609B1 (en) 2015-02-18
US20110149991A1 (en) 2011-06-23
EP2320609A1 (en) 2011-05-11
US8693472B2 (en) 2014-04-08
EP2320609A4 (en) 2011-08-17

Similar Documents

Publication Publication Date Title
WO2010020156A1 (zh) 一种混合业务流的缓存处理方法、存储转发方法及装置
JP5863076B2 (ja) パケットを再構築し再順序付けするための方法、装置、およびシステム
US8505013B2 (en) Reducing data read latency in a network communications processor architecture
CN103944837B (zh) 网络处理器单元和用于网络处理器单元的方法
CN108809854B (zh) 一种用于大流量网络处理的可重构芯片架构
EP0960536B1 (en) Queuing structure and method for prioritization of frames in a network switch
US6504846B1 (en) Method and apparatus for reclaiming buffers using a single buffer bit
US7787442B2 (en) Communication statistic information collection apparatus
US8514874B2 (en) Thread synchronization in a multi-thread network communications processor architecture
US5953335A (en) Method and apparatus for selectively discarding packets for blocked output queues in the network switch
CN108833299B (zh) 一种基于可重构交换芯片架构的大规模网络数据处理方法
JP4483535B2 (ja) ネットワーク装置
US6233244B1 (en) Method and apparatus for reclaiming buffers
CN100438481C (zh) 分组处理装置
JP4779955B2 (ja) パケット処理装置及びパケット処理方法
CN104641602B (zh) 输出实时网络流量延迟以及缓冲区占用
CN1638361A (zh) 网络交换设备的并行数据链路层控制器
US8868889B2 (en) Instruction breakpoints in a multi-core, multi-thread network communications processor architecture
EP0960505B1 (en) Method and apparatus for controlling initiation of transmission of data as a function of received data
US20110222552A1 (en) Thread synchronization in a multi-thread network communications processor architecture
US11151073B1 (en) Augmenting data plane functionality with field programmable integrated circuits
US9112708B1 (en) Processing multicast packets in a network device
JP2015511790A (ja) メッセージ処理方法及び装置
CN114531488B (zh) 一种面向以太网交换器的高效缓存管理系统
US8214553B2 (en) Virtualization of an input/output device for supporting multiple hosts and functions

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: 09807855

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 13059783

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 1535/DELNP/2011

Country of ref document: IN

WWE Wipo information: entry into national phase

Ref document number: 2009807855

Country of ref document: EP