JP4109693B2 - Node, RPR interface card and optical network system - Google Patents

Description

  The present invention relates to a packet ring interface circuit of an RPR node (node), selection of a shortest path, and selection of an optimum path in a packet ring to which, for example, an RPR (Resilient Packet Ring) protocol is applied. The present invention relates to a node, an RPR interface card, and an optical network system that are suitable for use in a network.

  Traffic such as Ethernet (Ethernet) packets and IP (Internet Protocol) packets transferred on the Internet increases, and the number of users (subscribers, general individuals, companies, etc.) using an ADSL (Asymmetric Digital Subscriber Line) network increases. The number of networks and the like using the VoIP (Voice over Internet Protocol) technology has risen rapidly.

(X1) Example of Optical Transmission System and Optical Network System FIG. 29 is a diagram illustrating a configuration example of an optical transmission system. The optical transmission system 500 shown in FIG. 29 is configured by connecting nodes (communication devices, RPR nodes, node devices, or ring nodes) provided in a wide area by optical fibers or the like, and is configured with a high-speed and large-capacity packet ( An IP packet, an IP datagram, or an IP frame), which is a backbone WDM (Wavelength Division Multiplexing) network 101, and an ADSL network 102a, SONET / SDH as a subnetwork connected to the WDM network 101 (Synchronous Optical NETwork / Synchronous Digital Hierarchy: Synchronous Optical Communication Network / Synchronous Digital Hierarchy) Network 102b, Internet 1 It is configured to include an RPR system 102d as 2c and the optical network system. Here, the WDM network 101 is a multiple packet ring configured by connecting a plurality of nodes by a plurality of optical fibers. The packet multiplexing method is, for example, time multiplexing, and a plurality of users are assigned to time slots. The ADSL network 102a is a transmission network that uses a telephone metallic cable and has a higher uplink transmission rate from the user to the telephone station than the reverse downlink transmission rate. The Internet 102c is used to multiplex and transmit large-capacity packets. In many cases, VoIP technology for transmitting and receiving voice data using the Internet 102c is used. For example, extension telephones and Internet telephones using LANs (Local Area Networks) or WANs (Wide Area Networks) of companies are used. It has been put into practical use.

  Furthermore, the SONET / SDH system 102b is configured such that each node (communication device) is connected via an optical fiber, and can transmit packets in both directions at high speed. The SONET / SDH system 102b provides a transmission service by allocating a band to each channel, and has a high reliability by switching channels in a short time when a failure occurs. On the other hand, the SONET / SDH system 102b may be inconvenient in bandwidth utilization efficiency, cannot manage optical fibers individually, and further, it is costly to expand the system as the traffic volume increases. It is extremely difficult from the technical and technical viewpoints.

For this reason, packet ring networks have attracted attention in recent years. In this packet ring, the ring itself processes the packet directly, manages the bandwidth of the packet traffic, and draws out the statistical multiplexing effect (effectively allocating network resources according to the packet traffic and distributing the load). Can do.
In particular, in urban areas such as metro areas, the tendency to introduce packet ring is remarkable. For this reason, IEEE (Institute of Electrical and Electronic Engineers) is working on standardization of IEEE 802.17 RPR (specifying the standard specification of the packet ring system) with the goal of December 2003. Conventionally, a method in which each node transmits and receives a packet using a token in a packet ring is known as a token ring or an FDDI (Fiber-Distributed Data Interface) ring. However, recent packet rings are distinguished from token rings and the like. Therefore, it is generally called an RPR network.

  The RPR network 102d shown in FIG. 29 is used for a bidirectional double ring with the WDM network 101, and is configured by connecting six nodes #A to #F in a ring shape. For example, the node #D Is connected to the packet ring and the Internet 102c. Here, the function of the node #D is (i) dropping the ring packet from the adjacent node #E or the node #C, converting it to, for example, an IP packet, and transmitting the IP packet to the Internet 102c. (Ii) The ring packet from node #E or node #C is passed through and relayed to node #C or node #E. (Iii) An IP packet from a server (not shown) on the Internet 102c is added, converted into a ring packet, and transmitted to the packet ring. (Iv) Transfer the IP packet from the server.

  Further, the RPR network 102d has high bandwidth utilization efficiency, and can be operated by inserting an RPR card (RPR interface card), which will be described later, into a device rack (not shown) provided in each of the nodes #A to #F. (This is called a plug and play function.) The layer supported by the RPR network 102d is a MAC (Media Access Control) layer, and various physical layers can be introduced.

(X2) Packet Format Example FIGS. 30A and 30B are diagrams showing examples of packet formats used in the RPR network, and represent MAC packets and RMAC (RPR MAC) packets as packet examples. This MAC packet is an Ethernet (registered trademark) packet defined in IEEE 802.3, and the RMAC packet has an RPR MAC header defined in IEEE 802.17. Here, the information written in each field will be schematically described. As is well known, the MAC packet shown in FIG. 30A is a MAC transmission destination address, a MAC transmission source address, an ether type, a payload (user data part), and an FCS (Frame Check Sequence) for error detection. Each field.

  Also, the Time To Live (hereinafter referred to as TTL) of the RPR header shown in FIG. 30B represents the packet active time, and the packet is “−1” every time one node is relayed, and the TTL is “ When it becomes “0”, it is discarded. RI (Ringlet Identifier) is an identification bit assigned to one of the bidirectional rings, and represents a ring that the packet originally transmits. For example, when a packet originally transmits an outer ring or an inner ring, it is set to 0 or 1, respectively. FE (Fairness Eligible) indicates whether the packet is subject to fairness control. PT (Packet Type) represents a packet attribute, for example, an RPR control packet, a data packet, or the like. SC (Service Class) is for identifying a class. WE (Wrap Enable) represents the presence or absence of a wrapping (wrapping) function. R is for reservation. A field for header error check bits is provided in front of the payload.

Thus, the RMAC packet covers each piece of information included in the MAC packet, and the IP packet and the ring packet are converted to each other.
(X3) Protection Function Using Wrap and Steering When a failure occurs in the RPR network 102d as an optical ring network, each node #A to #F performs a switching process (transfer route) of the received packet ( Wrapping processing).

(X4) Node of RPR Network 102d Each node #A to #F constituting the packet ring interfaces the packet ring which is the RPR network 102d and an external network (for example, the Internet 102c) different from the RPR network 102d. An RPR card is provided. There are two main types of hardware for this RPR card.

  FIG. 32 is a configuration diagram of a separation-type card provided in, for example, the node #D of the RPR network 102d. Each packet ring of the RPR network 102d generally forms a dual ring (double ring transmission path or double packet ring) composed of an outer ring 103a and an inner ring 103b in order to realize protection in wrapping and steering. Yes.

Also, the node #D shown in FIG. 32 has a configuration in which the East port and the West port are separated (hereinafter referred to as separate hardware), and the node #D shown in FIG. This is a configuration in which the West port is integrated (hereinafter referred to as integrated hardware).
(X5) RPR Card Node #D shown in FIG. 32 includes an East RPR card, a West RPR card, a switch unit (SW unit), and n (n represents a natural number) line cards. Is configured. In the separable card (East side RPR card and West side RPR card), the East side termination function and the West side termination function are provided in separate RPR cards, respectively. The switch unit switches the route between the packet on the RPR network 102d side and the packet on the external network side. The line card corresponds to various transmission specifications such as 10 Mbps, 100 Mbps, and 1 Gbps.

On the other hand, the integrated card shown in FIG. 33 is provided in the RPR card having the same termination functions on the East side and the West side.
(X6) Data Path in RPR Card FIG. 34 is a diagram showing a data path in a conventional separation-type RPR card, and FIG. 35 is a diagram showing a data path in a conventional integrated RPR card. The data path in the ring is displayed. Each of the RPR cards 200a and 200b shown in FIG. 34 and the RPR card 210 shown in FIG. 35 is a combination of dual rings of the outer ring 103a and the inner ring 103b shown in FIGS. 32 and 33, respectively. Yes. Ring packets that are not dropped in each of the node #D (FIG. 32) and the node #D (FIG. 33) are transferred between the RPR cards 200a and 200b and the east side and the west side via the RPR card 210. It is passed through and sent to the packet ring.

Accordingly, packets from the ring are not transmitted to the switch unit (SW unit) 7 side, but are processed in the RPR cards 200a and 200b and the RPR card 210, respectively. The route of this ring-through path is generally called a mate interface.
The following (Y1) and (Y2) show the characteristics of each configuration of the separation type card and the integrated type card.

(Y1) Features of the Separable Card The merit of the separate card is that maintenance replacement is possible for each RPR card (East card or West card) when a failure occurs in the RPR card. For example, when maintenance replacement becomes necessary due to a failure of the East card or the like, the West card is still mounted on the node (for example, node #D). For this reason, the wrap protection function operates, and communication can be continued without disconnecting the entire system from the ring.

Further, since the configuration of the RPR card is different when the conventional technology is used, there are the following differences in packet transfer between the line card and the RPR card.
(Y2) Features of the integrated card The advantage of the integrated card is that the node can be connected to the packet ring using an RPR card with one node. For this reason, the node provided with the integrated card can reduce the cost of the packet ring and the node as compared with the node provided with the separation type card. In addition, since it is possible to reduce the consumption of slots for accommodating RPR cards, it is possible to increase the number of ports of the entire node including the line cards.

  When the node uses an integrated card, the integrated RPR card has a ring topology table and the like, and has a destination ring selection function using this table. Therefore, the line card transmits the packet to the RPR card without special awareness of whether or not to transmit the packet to either the outer ring or the inner ring, and the RPR card uses the ring topology table. Packet was sent to the ring that was the shortest route. That is, the RPR card need only transmit a packet to the RPR card without paying particular attention to the East port or West port to which the packet is to be transmitted.

  In addition, a circuit driver / receiver device provided with two printed boards has been proposed for a circuit board (see, for example, Patent Document 1). This patent document 1 relates to a line driver / receiver device, and separates a line control unit into a line control print and a driver / receiver printed board for different interfaces, thereby enabling them to be shared. . This makes it very easy to change between different interfaces.

Further, a program controller is disclosed in which a bus conversion card having two or more connection channels with a PIO unit is mounted in a CPU unit (for example, see Patent Document 2). As a result, it is possible to easily change the duplication and non-duplication settings of the PIO interface unit and the PIO bus, and a large number of connectable PIO cards and systems that require high reliability are used according to the application. The required system can be freely constructed, and a programmable controller with a high degree of freedom can be obtained .

However, the separate card and the integrated card have the disadvantages shown in the following (Y3) and (Y4), and the problem shown in (Y5) when all RPR cards perform packet discard processing. .
(Y3) Integrated Card The disadvantage of the integrated card is that each node is completely disconnected from the packet ring when the RPR card is faulty or maintenance is replaced. For this reason, the integrated card is applied to a small-scale network or the like in which cost reduction is more important than reliability.

(Y4) Separable card There are mainly two disadvantages when using a separate card. First, since the RPR card is separated and each node #A to #F needs to mount at least two RPR cards, the cost for configuring the ring as compared with the nodes of the integrated card configuration It is a point to be up. The second point is that the number of ports allocated to the physical slot for inserting the RPR card (accommodating the RPR card) out of the total number of ports (port accommodation number) provided in the node is consumed, In addition, since the number of ports allocated to the physical slot for inserting the line card is also consumed, the total number of ports that can be used by the node is reduced. For this reason, the separation-type card is applied to a carrier network or the like owned by a communication carrier (communication carrier) who places importance on reliability rather than cost reduction.

Further, according to the prior art, due to the difference in the configuration of the RPR card, there are the following differences in packet transfer between the line card and the RPR card.
When the integrated card is used, when the line card transmits a packet to the RPR card, it is necessary to select the RPR card that is the shortest route. For this reason, the line card is provided with a destination ring selection function (ring topology table or the like), and the ring that is the shortest path is selected by this selection function. That is, the line card itself needs a function of selecting an East port or a West port to which a packet is to be transmitted and transmitting the packet to the RPR card connected to the selected port.

  On the other hand, when the separation type card is used, when the line card transmits a packet to the RPR card, it is necessary to select the RPR card that is the shortest route and transmit the packet. For this reason, the line card has a function (ring topology table or the like) for selecting a destination ring inside the line card, and selects the ring that is the shortest route. That is, the line card itself needs to have a function of determining the East port and West port to which a packet should be transmitted and transmitting the packet to an appropriate RPR card.

  (Y5) In addition to these methods, there is a method in which the line card transmits packets to two or more RPR cards, and the RPR card autonomously discards packets that are not the shortest route. Always needs to multicast packets to both or all RPR cards. For this reason, the line card, the switch (SW) card, and the RPR card transfer unnecessary packets, which increases the load on the packet processing unit 4 and causes a decrease in the packet processing capability of the node.

As described above, the packet transfer method between the line card and the RPR card in the conventional RPR system is originally different between the integrated card and the separated card, and thus the architecture of the packet transfer method of the node itself is different.
Therefore, conventionally, in the development of an RPR system, when either the integral type or the separated type is adopted as the configuration form of the RPR card, the developer can reduce the reliability and price for each user's packet ring. However, there is a problem that an optimal RPR system cannot always be provided.

The present invention has been devised in view of such problems, and in a node constituting a packet ring, communication can be continuously performed while disconnecting an RPR card, and costs can be reduced by reducing the number of interface cards. In addition, an object is to provide a node, an RPR interface card, and an optical network system capable of redundancy suitable for a network form when any configuration of interface card redundancy and ring redundancy is used.
JP-A-1-129554 Japanese Patent Laid-Open No. 2000-82018

  (1) For this reason, the node of the present invention is a node that switches a ring packet of a ring transmission path in which a plurality of nodes are connected via a multiple ring transmission path and a second packet of a plurality of second transmission paths. A node identifier that represents a source node of the ring packet, a ring identifier that represents each ring transmission path of the multiple ring transmission path, address information of the source apparatus of the second packet, and a connection state of a plurality of nodes. Based on the ring topology, based on the line card that transmits / receives the second packet to / from the second transmission path, the node identifier of the received ring packet, and the destination address of the second packet received by the line card A ring packet and a second packet are input / output to / from the line card side or the multiple ring transmission line side and connected to each other. A protocol processing unit, and a packet processing unit for switching each transfer path between a ring packet output from the plurality of protocol processing units to the line card and a second packet output from the line card to the plurality of protocol processing units. It is characterized by being composed.

Therefore, in this way, the RPR node can be provided with a separation type card and an integrated type card by the same RPR interface card. In any of these card configurations, redundancy, shortest path selection and It is possible to select the optimum route, and it is possible to select the shortest route and the optimum route even in the ring redundancy more than the quad ring.
Furthermore, the cost of the node can be reduced by reducing the number of parts. Communication interruption at the time of exchanging the RPR interface card is avoided, and continuous communication becomes possible. Furthermore, with regard to card manufacture and card installation, a large number of parts can be used for general purposes, the cost of the card can be reduced, and the cost of system construction can be reduced by reducing the number of RPR interface cards.

In addition, it is possible to ensure redundancy regardless of the configuration of the RPR interface card and the ring.
(2) A plurality of RPR interface cards provided with a plurality of protocol processing units and a packet processing unit are provided, and at least one of the plurality of RPR interface cards is connected to the first direction of the multiple ring transmission path. A first card connected to a port in the first direction among a plurality of ports provided in each of the second directions, and a second card connected to a port in the second direction among the plurality of ports It may be configured as a separate card provided separately, and in this way, the RPR interface card connected to the ring network is, for example, an East (first direction) port and, for example, a West (second direction) port. The packet ring system is disconnected from the ring network when the RPR interface card is replaced during maintenance operation. Rukoto no communication can be continued.

  (3) The RPR interface card described above may be configured as an integrated card connected to all of the plurality of ports provided in the first direction and the second direction of the multiple ring transmission path, respectively. For example, both the packet ring network with a highly reliable packet ring network and a packet ring network in which the East port and the West port are integrated and the cost reduction effect is high by reducing the number of RPR interface cards, etc. are used in common hardware. Can be realized.

  (4) A plurality of protocol processing units respectively include a first protocol processing unit that inputs and outputs a ring packet with respect to a part of the ring transmission lines of the multiple ring transmission lines, and a remaining part of the multiple ring transmission lines. A second protocol processing unit for inputting / outputting ring packets to / from the ring transmission path, and a first protocol processing unit provided in the same RPR interface card among the plurality of RPR interface cards and a second protocol processing unit The two protocol processing units are connected via the internal mate interface, and the first protocol processing unit and the second protocol processing unit provided in separate RPR interface cards of the plurality of RPR interface cards are the external mate interface. May be configured through a connection, and in this way a separate car Whichever is used with the integrated card and can share a switch unit and the line cards, also cost can be reduced.

(5) a plurality of protocol processing units, respectively, based on the destination address of the ring packet, determines whether the ring packet is a to some other node destined addressed to its own node, if it is determined that the own node addressed is while transfers the ring packet to the packet processing unit, if it is determined that another node addressed, can also configure the ring packet to through the multiple ring transmission path, in this manner, during maintenance operations The communication can be continued without disconnecting the packet ring system from the ring network when the card is replaced, the reliability of the system is improved, and the cost can be reduced by reducing the number of installed RPR interface cards.

  (6) Furthermore, a hash calculation result holding unit that holds correspondence data in which the calculation result of the hash calculation is associated with the ring identifier is provided, and the correspondence held in the hash calculation result holding unit when a failure occurs in the multiple ring transmission path By updating the data, it may be configured to stop broadcast packet distribution via the failed ring transmission line and continue load distribution of broadcast packet distribution. Broadcast packet distribution to the ring stops and load distribution of broadcast packet distribution continues.

  (7) According to the RPR interface card of the present invention, each of the ring packet and the second packet is divided into a plurality of ring packets and a second packet based on the node identifiers of the received ring packets and the destination addresses of the second packets received from the plurality of second transmission paths. A plurality of protocol processors connected to each other and connected to the second transmission line side or the multiple ring transmission line side; a ring packet output from the plurality of protocol processors to the plurality of second transmission lines; And a packet processing unit that switches each transfer route with the second packet output from the two transmission lines to the plurality of protocol processing units, and is provided in each of the first direction and the second direction of the multiple ring transmission line. A first card connected to a port in the first direction among the plurality of ports and a second card connected to a port in the second direction among the plurality of ports. It is characterized in that the over de configured as a separate-type card separately provided.

Therefore, with this configuration, a redundancy function suitable for the network topology can be obtained in both the redundant configuration and the ring redundant configuration of the RPR interface card. Furthermore, since each node dynamically manages the ring topology table, it is possible to always transfer a packet on the shortest path in the packet ring.
(8) Further, the card of the present invention can receive a plurality of ring packets and a plurality of second packets based on node identifiers of received ring packets and destination addresses of second packets received from a plurality of second transmission paths. A protocol processor connected to and input from the second transmission line side or the multiple ring transmission line side; a ring packet output from the protocol processing part to the plurality of second transmission lines; and a plurality of second transmission lines And a packet processing unit that switches each transfer route with the second packet output to the protocol processing unit side, and is connected to all of the plurality of ports respectively provided in the first direction and the second direction of the multiplex ring transmission line It is characterized by being configured as an integrated card.

Therefore, in this way, an integrated card and a separate card can be used, and each node can be manufactured using common hardware, and further each of the RPR interface card redundancy and the ring redundancy. A redundant function suitable for the network form can be obtained.
(9) A ring in which the line card holds a node topology, a ring identifier, and a learning table that holds the address information of the transmission source device of the second packet in association with each other, and a ring topology that represents a connection state of a plurality of nodes. A ring packet transferred via the shortest path based on the node identifier of the source node of the broadcast packet and the ring topology, when the learning table has a quadruple or more multi-ring transmission line. Can be configured to learn the shortest path for transferring the broadcast packet by discarding the broadcast packet, and in this way, the packet that becomes the shortest path in the learning table of the line card can be configured. It is possible to learn about the ring and transmit efficiently. .

  (10) In the optical network system of the present invention, at least one of the plurality of nodes or the ring transmission path is received from the node identifier of the received ring packet and the second transmission path. A plurality of protocol processing units connected to each other by inputting / outputting ring packets and second packets to / from a plurality of second transmission line sides or multiple ring transmission line sides based on packet destination addresses, and a plurality of protocols A packet processing unit for switching each transfer path between the ring packet output from the processing unit to the plurality of second transmission paths and the second packet output from the plurality of second transmission paths to the plurality of protocol processing units; A plurality of ports provided in each of the first direction and the second direction of the multi-ring transmission line. A separation-type card in which a first card connected to the first-direction port and a second card connected to the second-direction port among the plurality of ports are separately provided; and a multiplex ring transmission line The integrated card connected to all of the plurality of ports respectively provided in the first direction and the second direction is mixed.

Therefore, in this way, a common processing method can be used for packet processing in each node, and line cards and switch cards can be shared. Further, each node can perform card redundancy, shortest path selection, and optimum path selection. Further, the shortest path selection and the optimum path selection are possible even in a quadruple or more packet ring .

(A) Description of First Embodiment of the Present Invention FIG. 1 is a configuration diagram of an optical transmission system according to a first embodiment of the present invention. An optical transmission system 100 shown in FIG. 1 is a high-speed and high-bandwidth network system in which a wide area is connected by an optical fiber or the like, and includes a WDM network 101, an RPR network (optical network system) 102, The network 102a-102c is comprised.

  Here, the WDM network 101 is configured by connecting nodes provided in a wide area (for example, all over North America) by an optical fiber or the like, and transmits high-speed and large-capacity packets. It functions as a network. Further, each of the sub-networks 102a to 102c has a network protocol of various specifications such as an ADSL network, a SONET / SDH network, and the Internet.

(1) RPR network 102
The RPR network 102 is an optical network system to which the present invention is applied. For example, the RPR network 102 transmits packets between large cities. The RPR network 102 is connected to a plurality of LANs such as Gigabit Ethernet in each large city, and provides a user with an environment such as MAN and WAN. By using the RPR network 102, the user is provided with services such as convenience and utilization efficiency of the local network.

Further, an external network 107 (described later) such as the Internet is connected to the RPR network 102.
(1-1) Configuration of Multiple Packet Ring (Multiple Ring Transmission Path. Multiple RPR Ring or Multiple Ring Network) FIG. 2 is a schematic configuration diagram of the RPR network 102 according to the first embodiment of the present invention. In the RPR network 102 shown in FIG. 2, for example, six nodes (RPR nodes or ring nodes) #A to #F are connected in a ring shape using duplex optical fibers (a pair of optical fibers). A dual ring (double ring transmission line) is formed.

(1-2) External network 107
The one or more external networks 107 connected to the nodes #A to #F are, for example, Ethernet, Internet, LAN, WAN, or MPLS (Multi-protocol Label Switching) network (or user line, line network). Note that MPLS is a technology in which a router (not shown) in an MPLS network assigns a label to an incoming packet instead of an IP address, and each router provided in the MPLS network transfers an MPLS packet by referring only to the label. It is.

  As a result, each of the nodes #A to #F terminates the ring packet received from the packet ring, converts the terminated packet into a packet of the format in the external network 107, and converts the converted packet to the external network 107. Send. Each node #A to #F terminates a packet received from the external network 107, converts it into a ring packet format, and transmits it. Therefore, each node #A to #F interfaces the packet in the packet ring with the packet in the external network 107, and switches the packet.

That is, each of the nodes #A to #F can switch the packet regardless of the interface type, and can switch or interface the packet without depending on the physical layer type.
In the following description, the external network 107 will be described as the Internet unless otherwise specified. Further, the RPR network 102 can process a MAC layer packet without depending on a physical layer protocol.

The RPR network 102 shown in FIG. 1 is connected to a plurality of external networks (Ethernet, Internet, LAN, WAN, or MPLS network) 107, and these Ethernet, Internet, LAN, WAN, or MPLS networks are processed or transferred. Packets function as second packets of a plurality of second transmission paths.
(1-3) Ring Multiplexing and Direction Descriptions FIGS. 31 (a) to 31 (c) are diagrams for explaining multiple ring transmission paths (multiple packet rings), respectively. What is indicated by “shape” represents one optical fiber. In the packet ring shown in FIG. 31A, the outer ring # 1 (Tx, Rx) and the inner ring # 2 (Tx, Rx) are configured as a dual ring. That is, one ring consists of two optical fibers. Each node #A to #F can dynamically allocate a packet transmission band for each ring. Tx and Rx represent transmission and reception, respectively.

  Also, the outer (first direction) optical fiber shown in FIG. 2 corresponds to the outer ring (outer packet ring), and the inner (second direction) optical fiber corresponds to the inner ring (inner packet ring). This number of ring multiplexes can be expanded to triple (see FIG. 15A described later) and quadruple as shown in FIGS. 31 (b) and 31 (c), respectively. That is, each node #A to #F is connected via a double to quadruple packet ring. Thereby, the amount of packets can be expanded according to the increase in traffic.

Further, in the following description, the node #F side is referred to as the West side and the node #B side is referred to as the East side when viewed from the node #A. Nodes #B to #F are similar to node #A. West and East can also be defined in the opposite direction.
(1-4) Redundant configuration (protection)
The packet ring is made redundant by this dual ring, and the outer ring # 1 and the inner ring # 2 each function as an active one. Here, when a failure occurs, each of the nodes #A to #F activates wrap protection and selects a packet path so as to select a normal route in order to avoid packet transmission to the failure ring. Note that various changes can be made as to whether the packet is transmitted to either the outer ring # 1 or the inner ring # 2.

The triple and quadruple packet rings are switched in the same manner as the dual ring switching. With these redundant configurations, the recovery time at the time of failure is relatively short, and the transmission distance is shortened.
(1-5) Features of Multiple Packet Ring Each node #A to #F has an RPR card (RPR interface card, interface card or IF [Interface] card or card) described in detail below. A packet ring is connected to an external network 107 such as the Internet. Since each of the nodes #A to #F is not connected to a network different from both the packet ring and the external network 107, the number of ports that need to be installed can be reduced, and the cost can be reduced by reducing the number of ports.

Further, since the RPR card of each node #A to #F has a plug-and-play function, the number of nodes in the RPR network 102 can be easily increased and decreased, and the trouble of setting the line by the administrator (operator) is reduced. The maintenance management of the RPR network 102 is simplified.
In addition, the RPR network 102 has a statistical multiplexing function. A definition example of this statistical multiplexing function is to efficiently allocate network resources according to packet traffic and distribute the load.

(2) Configuration of Nodes with Separable Cards Each node #A to #F is described below as a ring packet of a packet ring in which six nodes #A to #F are connected via a multiple packet ring. This switches the external network 107 connected via n (n represents a natural number) line cards 8. The RPR cards provided in the nodes #A to #F are each configured as a separation type card provided with two RPR cards.

  FIG. 3 is a block diagram of the node #A provided with the separation type card according to the first embodiment of the present invention. The node #A shown in FIG. 3 transmits / receives a ring packet to / from the nodes #B and #F via a quad ring in which packet rings (rings) # 1 to # 4 are multiplexed four times. A card 1a, a second RPR card 1b, n line cards 8, a switch unit (SW unit) 7, a CPU bus (bus line) 12a, and a CPU unit 12 are provided. In addition, since all or part of the nodes #B to #F have the same configuration as the node #A, redundant description of the nodes #B to #F is omitted.

(2-1-1) RPR cards 1a and 1b
Both RPR cards 1a and 1b are provided with two RPR units (protocol processing units) 3a and 3b and a packet processing unit 4.
These RPR cards 1a and 1b are connected to both rings # 1 and # 2 and rings # 3 and # 4, respectively, and ring packets from rings # 1 to # 4 and line cards 8 Switch or interface with IP packets. Here, the RPR unit 3a of the first RPR card 1a and the RPR unit 3a of the second RPR card 1b are connected to each other via a mate interface (external mate interface 6 to be described later). This is different from the RPR card with a mate interface.

The quad ring is configured by two pairs of dual rings. When only rings # 1 and # 2 are used for transmission, rings # 3 and # 4 can be unused.
The configuration of the RPR cards 1a and 1b will be described later.
(2-1-2) Line card 8
Each of the n line cards 8 includes a node ID (node identifier) that represents a transmission source node of the ring packet, a ring ID (ring identifier) that represents each packet ring of a multiple packet ring such as double to triple, An IP packet is transmitted / received to / from the external network 107 based on address information such as a MAC address of an IP packet transmission source device and a ring topology representing connection states of six nodes #A to #F. is there.

Each line card 8 is also connected to the external network 107, and terminates the protocol of the external network 107 and the RPR protocol, and switches or interfaces the ring packet and the IP packet.
Note that the value of n can be changed according to the increase or decrease of subscribers. That is, the number is increased or decreased based on the number of lines and the number of users that can be accommodated (or multiplexed) by one line card 8.

  Each line card 8 is, for example, a circuit board (switch card), and includes circuit components, circuit patterns, connection terminals, and the like, as well as a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access Memory). ), IC (Integrated Circuit), LSI (Large Scale Integration), and other chips or devices are provided, and these perform the interface function.

The configuration of the line card 8 will also be described later.
(2-1-3) Switch unit 7
The switch unit 7 has a packet switch (interface or exchange) function in packet transfer between the RPR cards 1a and 1b and the n line cards 8, and is realized by a switch card, for example.

The switch unit 7 is not necessarily provided in the nodes #A to #F where the RPR cards 1a and 1b and the line card 8 are directly connected via a backboard (not shown). In addition, the number of switch units 7 can be increased or decreased, and each node #A to #F can be provided with a desired number of switch cards.
(2-1-4) CPU unit 12
The CPU unit 12 controls the entire apparatus of the node #A, and controls each interface card and line card 8 via the CPU bus 12a of the data path provided in the node #A. Specifically, each RPR card of separation type or integral type is set via the CPU bus 12a. The function of the CPU unit 12 is realized by, for example, a CPU (Central Processing Unit), ROM, RAM (not shown), and the like, and is formed as a CPU card in which the CPU, ROM, RAM, etc. are provided on one circuit board. ing.

The node #A can be provided with two or more CPU cards, or two or more CPUs can be provided on one CPU card. The number of CPU cards or the number of CPUs can be increased or decreased, and the processing capacity can be expanded or reduced according to the load.
(2-1-5) Schematic description of packet transmission As a result, a packet from the external network 107 connected to the node #A shown in FIG. 1 to the external network 107 connected to the node #B is transmitted to the node #A. After being formatted in the line card 8, the switch unit 7 of the node #A switches to the RPR card 1a or RPR card 1b of the node #A selected according to the packet transmission destination. The switched packet is converted into a format for transmitting the packet ring and transmitted to the ring.

When the node #B receives the packet, it converts the packet format, inputs the converted IP packet, for example, to the desired line card 8 via the switch unit 7 of the node #B, and connects to the external network connected to the node #B. It transmits to 107. The same applies to packets switched to the RPR card 1b.
(3) RPR cards 1a and 1b
FIG. 4 is a block diagram of the RPR cards 1a and 1b according to the first embodiment of the present invention, and the RPR cards 1a and 1b have the same configuration. The RPR card 1a shown in FIG. 4 is connected to each of the node #F on the West side and the node #B on the East side via ports and optical fibers and to one or more external networks 107. The RPR packet in the packet ring is interfaced by using the through function and the switch function described below.

(3-1) Through, Switch, and Return Using the above packet transfer route, the RPR card exhibits the through, switch, and return functions.
Through means that a packet that is not dropped at a node is transmitted to the packet ring. That is, a packet having a transmission destination of a node other than its own node among packets from the East side or the West side is transmitted again to the West side or the East side via the internal mate interface 5 or the external mate interface 6.

  Specifically, for example, the node #A receives the ring packet from the adjacent node #F recognized by the topology data held in the ring topology table (see FIG. 7) described later based on the transmission destination information of the ring packet. While transmitting to the other adjacent node #B, the MAC packet from the external network 107 is transmitted to the external network 107.

Furthermore, the switch converts the format of the packet from the external network 107, switches the route to two or four optical fibers based on the transmission destination of the packet, and formats the packet from the packet ring. This means that the route is switched to the external network 107 based on the destination of the packet after conversion.
The return is a protection operation (protection function) when a failure occurs in the optical fiber connecting each node or each of the nodes #A to #F. In order to prevent the received packet from being discarded or lost, The received packet is returned to the other ring different from the receiving ring. Specifically, during the protection operation, a packet received from one of the outer ring 103a and the inner ring 103b is transferred again to the other inner ring 103b or outer ring 103a in the RPR card 1.

Thereby, the RPR unit 3a transmits, for example, a packet from the packet ring # 1Rx to the adjacent RPR card 3a through the internal mate interface 5, and a packet from the packet ring # 1Rx at the time of failure. Is transmitted again to the packet ring # 1Tx. That is, it is used for passing and returning packets.
Each node #A to #F is provided with a mate interface management unit 4a that can exclusively control the internal mate interface 5 and the external mate interface 6, and the internal mate is managed by the exclusive control management of the mate interface management unit 4a. It is prohibited to energize (or operate) both the interface 5 and the external mate interface 6 at the same time. The installation location of the mate interface management unit 4a may be provided outside the RPR units 3a and 3b, the packet processing unit 4 or the RPR cards 1a and 1b. A mate interface management unit 4 a shown in FIG. 4 is provided in the packet processing unit 4.

Note that whether the internal mate interface 5 or the external mate interface 6 is used can be selected by setting of the administrator.
As a result, packets from the packet ring are not transmitted to the switch side, but are processed by the RPR card 1 via the route of this ring-through path. Both the internal mate interface 5 and the external mate interface 6 are referred to as both mate interfaces.

(3-2) Configuration of RPR Cards 1a and 1b The RPR cards 1a and 1b are, for example, card-type circuit boards, which are circuit components, circuit patterns, connection terminals, chips such as CPU, ROM, RAM, IC and LSI, or Devices are provided, and these perform the switching function.
The RPR cards 1a and 1b include two PHY units 2a and 2b, two RPR units 3a and 3b, a packet processing unit 4, an internal mate interface 5, and one or more external mate interfaces 6. Configured. The RPR cards 1a and 1b are connected to a double packet ring including an outer ring 103a and an inner ring 103b.

(3-3) PHY part 2a, 2b
Each PHY unit 2a, 2b controls a physical layer. The physical layers that can be processed by the RPR cards 1a and 1b of the present invention are various protocols. It should be noted that the functions of the PHY units 2a and 2b are both exhibited by an LSI (Large Scale Integration) or an FPGA (Field Programmable Gate Array).

(3-4) RPR portions 3a and 3b
Each RPR unit 3a, 3b receives the ring packet and the IP packet based on the node ID of the received ring packet and the destination address of the IP packet received by the line card 8, respectively. Are connected to each other. The RPR units 3a and 3b perform RPR layer processing and are provided with the same components.

The RPR units 3a and 3b can also be configured by LSIs or FPGAs. In this way, the functions of the two RPR units 3a and 3b can be physically used by using one LSI or FPGA. It is feasible.
FIG. 5 is a view for explaining switch processing in the separation type card configuration according to the first embodiment of the present invention. The RPR unit 3a of the RPR card 1a shown in FIG. 5 and the RPR unit 3a of the RPR card 1b are respectively provided in the node #A, and are connected to each other by the external mate interface 6 inside the node #A. It has come to be. In addition, the RPR unit 3b of the RPR card 1a and the RPR unit 3b of the RPR card 1b are connected using the external mate interface 6, respectively. Also, the nodes #B to #F have the same configuration as the node #A shown in FIG.

As a result, when packet rings are input to the node #A from the four packet rings # 1 to # 4, the RPR cards 1a and 1b and the external mate interface 6 provided in the node #A are respectively connected. Is passed through.
Specifically, when the RPR card 1a receives a packet from the ring # 1 (Ring # 1 Rx), the RPR unit 3a extracts the node ID of the destination node #B, for example, and extracts the packet. Whether the node ID is the node #A itself or another node #B is determined. When the packet destination node is the node #B, the RPR unit 3a is input to the RPR unit 3a provided in the RPR card 1b via the external mate interface 6. Then, the packet is switched by the RPR unit 3a, transmitted to the ring # 1, and passed through.

Even when a packet is input from the ring # 2 to the node #A, the RPR card 1a performs the same process as the process input to the node #A, and thus a duplicate description is omitted.
Further, when a ring packet from ring # 3 is input to node #A, RPR unit 3b provided in RPR card 1a refers to the header of the packet and transmits the packet to external mate interface 6. The packet input to the RPR unit 3b of the RPR card 1b and input to the RPR unit 3b is transmitted again to the ring # 3, thereby exhibiting a through function.

In addition, when a packet from the ring # 4 is input to the node #A, the RPR unit 3b of the RPR card 1b is input to the RPR unit 3b of the RPR card 1a via the external mate interface 6, and this RPR unit 3b is transmitted to ring # 4.
In other words, the RPR units 3a and 3b of the node #A determine their own node / other nodes based on the address information of the received packets from the PHY units 2a and 2b, respectively, and determine that they are addressed to their own node #A In this case, the ring packet is returned to another ring different from the received packet ring of the ring packet, and when it is determined that it is destined for another node, the ring packet is subjected to through processing to the multiple packet ring.

As described above, even when a packet is received from any one of the optical fibers in the dual ring, the node #A can accurately recognize each transmission destination by referring to the header of each packet, and the switch or interface. Can be done reliably.
On the other hand, also in the ring direction from the external network 107 side, the RPR units 3a and 3b transmit the received packets from the packet processing unit 4 to the PHY units 2a and 2b, respectively.

(3-5) Packet processing unit 4
For example, the packet processing unit 4 transfers each of the ring packet output from the two RPR units 3a and 3b to the line card 8 and the IP packet output from the line card 8 to the two RPR units 3a and 3b. The path is switched.
The packet processing unit 4 is configured to perform termination processing such as RPR packet generation, disassembly, and header replacement, and a packet generation function for transmitting ring packets to the packet ring and received from the packet ring. A ring packet has a decomposition (decompression, extraction) function.

  Specifically, for example, the packet processing unit 4 in the node #A that transmits the ring packet writes the address information of the transmission source device of the IP packet and the node ID in the ring packet. On the other hand, the packet processing unit 4 switches the transfer path of the ring packet based on the node ID, the ring ID, the address information of the IP packet transmission source device, and the ring topology, and between the integrated card and the separation type card. A common packet transfer process is performed.

The packet processing unit 4 includes, for example, a memory (not shown) that holds a packet, a CPU, a ROM, a RAM, an IC, and an LSI.
(3-6) Internal mate interface 5
The hardware of the internal mate interface 5 includes a circuit pattern of the RPR card 1, connection terminals, and conductive members such as wire wires (hereinafter referred to as wiring). The internal mate interface 5 connects the two RPR units 3a and 3b of the RPR cards 1a and 1b to each other.

(3-7) External mate interface 6
One or a plurality of external mate interfaces 6 are connected to the RPR unit 3a and RPR unit 3b of other adjacent RPR cards 1a and 1b, respectively.
The two RPR units 3a and 3b each receive and output a ring packet between a part of the packet rings of the multiple packet ring and the remaining packets of the multiple packet ring. A first RPR unit 3a and a second RPR unit 3b provided in the same RPR card 1 of the two RPR cards 1; Are connected via the internal mate interface 5 and the first RPR unit 3a and the second RPR unit 3b provided on separate RPR cards 1 of the two RPR cards 1 are connected via the external mate interface 6. Has been.

(3-8) Description of Packet Transfer Route Next, for RPR card 1, between East side external ring # 1 (Rx) and ring # 2 (Tx) and West side ring # 2 (Rx) and ring Each between 1 # (Tx) will be described.
A packet from the external ring # 1 (Rx) on the east side is input to the RPR unit 3a via the PHY unit 2a, and whether the destination of the received packet is the local node or another node in the RPR unit 3a. Is determined. When the packet is the local node, the RPR unit 3a acquires the packet. When the packet is another node, the packet is input to the RPR unit 3a of the adjacent RPR card.

Further, when the RPR unit 3a receives a packet from the RPR unit 3a on the east side of the adjacent RPR card, the RPR unit 3a determines the transmission destination (local node or other node) of the packet, acquires the packet addressed to the local node, and A packet from the RPR card to be transmitted is transmitted to the outer ring # 2 (Tx).
The same applies to the transfer on the West side.

  A packet from the external ring # 2 (Rx) on the West side is input to the RPR unit 3b via the PHY unit 2b, and whether the destination of the received packet is the local node or another node in the RPR unit 3b. Is determined. When the packet is the local node, the RPR unit 3b acquires the packet. When the packet is another node, the packet is input to the RPR unit 3b of the adjacent RPR card.

Further, when the RPR unit 3b receives a packet from the RPR unit 3b of the adjacent RPR card, the RPR unit 3b determines a transmission destination (local node or another node) of the packet, acquires a packet addressed to the local node, and adjacent RPR card To the outer ring # 1 (Tx).
Accordingly, each external mate interface 6 is connected to the outside of the RPR card 1.

(4) Line card 8
The line card 8 includes a learning table 10, a packet processing unit 9, and a ring topology table 11.
Here, the packet processing unit 9 controls packet transfer processing in the line card 8.

  The learning table 10 holds node IDs, ring IDs, and address information of IP packet transmission source devices in association with each other. The learning table 10 represents the source address (for example, the MAC address of the source node) indicating the client information of the packet and the RPR packet information in the packet transfer in the direction from the RPR card to the line card 8 (hereinafter referred to as “Ingress direction”). Learning by linking the node ID and the ring ID. In packet transfer from the line card 8 to the RPR card direction (hereinafter referred to as “Egress direction”), the learning table 10 is searched from the transmission destination address (for example, transmission destination MAC address) which is the client information of the packet, and the learned node ID And take out the ring ID.

The ring topology table 11 holds a ring topology representing the connection state of the six nodes #A to #F, and the ring topology state is always managed.
(5) Packet Format FIGS. 6 (a) to 6 (c) are diagrams showing a packet format according to the first embodiment of the present invention. Here, FIG. 6A to FIG. 6C represent a client packet, a device internal packet, and an RPR bucket, respectively, and the relative relationship of each packet format is displayed.

  The client packet shown in FIG. 6A is a packet input / output between the line card 8 and the external network 107, and is a normal Ethernet packet, IP packet, MPLS packet, or the like. Note that MPLS is a technology in which a router assigns a label instead of an IP address to various client packets, and each router refers to only the label and transfers the packet.

  6B is used inside each of the nodes #A to #F, and is transmitted / received between the RPR card and the line card 8. Here, the apparatus internal header stores packet information required between the RPR card and the line card 8. The transmission destination card bitmap is a field processed by the switch unit 7 and is assigned bits corresponding to the number of interface cards to be accommodated. Multicast transfer is possible by the bitmap format.

  For example, when eight interface cards can be installed, the field of the transmission destination bitmap is 8 bits, and each bit indicates on or off of the assigned interface card transmission destination. The ring ID and node ID indicate the source ring ID and source node ID of the packet in the direction from the RPR card to the line card 8, and the destination ring ID and destination node ID in the direction from the line card 8 to the RPR card. Show.

Accordingly, in the direction from the RPR card to the line card 8, the node ID stored in the source address of the RPR packet is stored in the node ID field of the device internal packet, and in the direction from the line card 8 to the RPR card, the device internal packet The node ID is stored in the transmission destination address field of the RPR packet.
In addition, E and W shown in FIG. 6B represent an East bit and a West bit, respectively. For the ring indicated by the ring ID, the packet is transferred via the East route and transferred via the West route. Identify. In addition, a bitmap format independent of E and W is used so that transmission is possible on both the East side and West side paths. If both E and W bits are on, the packet is sent to both East and West rings.

  TTL-East and TTL-West are active times used when packets are transmitted to the East route and West route, respectively. In the direction from the line card 8 to the RPR card, the TTL-East and TTL-West of the device internal packet (FIG. 6B) are inserted into the TTL field of the RPR packet (FIG. 6C) according to the packet transmission path. .

CRC information and the like indicating the normality of the packet are stored in the CRC and FCS.
Further, the RPR packet shown in FIG. 6C has a packet format transmitted and received in the packet ring. RPR packet information is stored in the RPR packet header. The destination address stores a node ID to which the packet should reach, and stores information indicating broadcast at the time of broadcasting. Normally, for example, all “1” is inserted.

The transmission source address represents a transmission source address that is client information of the packet. For example, transmission source MAC addresses of various servers (server devices) connected to the Internet are written.
TTL stores the active time of the packet. For the broadcast packet, the packet processing unit 4 changes the transfer path of the broadcast packet by setting the value representing the active time of the ring packet to a desired value.

CoS (Class of Service) indicates a packet class. The ToS field of the IP address, the priority of the VLAN (Virtual LAN) tag, the experimental use of MPLS, and the like are stored according to the policy of the optical network system.
Type indicates the type of the RPR packet, and the data stored in the RPR packet represents each type of information data, control data, ring topology table update data, protection data, or the like. HEC (Header Error Check) indicates the normality of the RPR packet header, and stores CRC bits and parity bits. CRC information amount indicating the normality of the packet is stored in the CRC and FCS.

In this way, by converting the internal packet of the apparatus shown in FIG. 6B into the format shown in FIG. 6C, the same packet forwarding process can be performed inside the apparatus regardless of whether it is an integrated type or a separated type.
Note that these pieces of information are information necessary for the present invention, and the header information of the device internal packet and the RPR packet is other information in addition to the above contents depending on the system to be applied and the RPR protocol to be applied. Can also be stored.

(6) Ring topology table 11
FIG. 7 is a diagram for explaining the ring topology table 11 according to the first embodiment of the present invention. The ring topology table 11 shown in FIG. 7 represents the ring topology table 11 of the node #A shown in FIG. 2, and includes items of ring ID, node ID, number of East hops, number of West hops, East route selection, and West route selection. Have.

Here, the ring ID is ring identification information and represents a physical ring of the dual ring or quad ring shown in FIGS. 31 (a) to 31 (c). Further, since the ring topology table 11 is a dual ring, the nodes #A to #F all store the ring ID # 1.
The node ID is identification information of each node #A to #F existing in the packet ring, and each node #A to #F inserts an ID (#A to ##) inserted into a topology message transmitted and received to generate topology data. F) is stored.

  The number of East hops and the number of West hops both represent the number of hops from the node #A to the other nodes #B to #F with the node #A as a reference (starting point), and each hop number is calculated from the TTL of the topology message. The For example, the number of East hops and the number of West hops of node #B are 1, 5 respectively. This “1” indicates that the number of hops to the node #B counted clockwise from the east side of the node #A is 1, and “5” is counted counterclockwise from the west side of the node #A. This indicates that the number of hops to node #B is 5. The number of East hops and West hops of node #A is 0.

  Both the East route selection and the West route selection represent a route through which the node #A transmits a packet. That is, based on the number of East hops and the number of West hops, it is set whether or not a packet from the node #A to each of the nodes #B to #F should be transmitted to East and West. For example, for the node #B, since the East side is close, the East route selection is set to ON (ON), and the West route selection is set to OFF (OFF). Here, ON represents selection and OFF represents non-selection.

Therefore, the ring topology table 11 selects a route based on the shortest route. Note that various methods can be used as a criterion for route selection.
That is, the packet processing unit 4 switches the transfer path of the ring packet based on the node ID, the ring ID, the address information of the IP packet transmission source device, and the ring topology, and between the integrated card and the separated card A common packet transfer process is performed.

Thereby, each node broadcasts a topology message to the packet ring periodically or when a failure occurs. The topology message normally stores its own node ID and TTL (packet active time) maximum value. Each of the nodes #A to #F receives this topology message and generates a ring topology table 11.
In this way, each of the nodes #A to #F dynamically manages the ring topology table 11, so that packets can always be transferred through the shortest path in the packet ring.

  The RPR network (optical network system) 102 according to the present invention includes a packet ring ring packet in which, for example, six nodes #A to #F are connected via a multiple packet ring, and a plurality of external network 107 IP packets. Nodes #A to #F for switching between are provided. Then, one or more nodes #A to #F among the above six nodes #A to #F receive the node IDs of the received ring packets and the destinations of the IP packets received from the plurality of external networks 107. Two RPR units 3a and 3b and two RPR units connected to each other by inputting / outputting ring packets and IP packets to / from a plurality of external network 107 sides or multiple packet ring sides based on addresses Packet processing unit 4 for switching each transfer path between the ring packet output from 3a, 3b to the plurality of external networks 107 and the IP packet output from the plurality of external networks 107 to the two RPR units 3a, 3b. Are provided with two RPR cards 1. The RPR network 102 includes a first RPR card connected to a half of the m ports (m represents a natural number) provided on the outer and inner sides of the multiple packet ring, and m A separate card that is provided separately from the second RPR card connected to the remaining half of the ports, and all m ports that are provided on the outer and inner sides of the multiple packet ring. The integrated card is mixed.

As a result, the interface card connected to the packet ring is separated into the East port and the West port, and when replacing the RPR card at the time of maintenance operation, any of the nodes #A to ## in which the RPR card has been replaced Communication can be continued without disconnecting F from the packet ring.
Furthermore, according to the RPR card 1 of the present invention, a highly reliable packet ring network, an East port, and a West port are integrated, and a packet ring with a high cost reduction effect due to a reduction in the number of interface cards is shared hardware. Can be provided at.

In addition, according to the RPR card 1 of the present invention, a redundant function suitable for the network topology is provided in both the redundant configuration of the interface card and the ring redundant configuration.
(7) Explanation of operation of node provided with separation type card With this configuration, for example, the operation of node #A will be explained. Since the operation of the nodes #B to F is the same as the operation of the node #A, a duplicate description is omitted.

  When the node #A receives a packet from the packet ring, the received packet is transferred from the PHY units 2a and 2b to the RPR units 3a and 3b. The RPR units 3a and 3b check the destination address of the received packet. If the destination address is addressed to the own node, the RPR unit 3a, 3b transfers the received packet to the packet processing unit 4, and if not addressed to the own node The packet is passed through the external mate interface 6 and transmitted to the ring again.

In the case of a broadcast packet, the node #A captures the packet in its own node and transmits the packet to the ring again using the external mate interface 6. The packet processing unit 4 removes the header of the received packet and transmits it to the switch unit 7.
At this time, the node #A assigns the node ID of the transmission source node that is the packet transmission source on the ring and the ring ID indicating which packet ring (Ring) # 1 to 4 received from the switch to the switch Send to part 7. When the line card 8 receives a packet from the switch unit 7, the packet processing unit 9 learns the client transmission source address (for example, transmission source MAC address), the node ID, and the ring ID of the packet in the learning table 10.

  Next, when the node #A transmits a packet to the packet ring, it is as follows. When the packet processing unit 9 of the line card 8 receives a packet from the outside, the packet processing unit 9 searches the learning table 10 based on the client transmission destination address (for example, transmission destination MAC address) of the packet. Then, the ring ID and the node ID are extracted from the learning table 10 by searching the learning table 10.

  The packet processing unit 9 searches the ring topology table 11 based on the ring ID and the node ID extracted by searching the learning table 10. Thereby, the packet processing unit 9 determines whether or not to transmit a packet to the RPR card 1 of either East or West in packet transmission to the desired nodes #B to #F and the East route selection bit. This is determined using the West route selection bit. Then, the packet processing unit 9 transmits the packet to the switch unit 7 based on each information of the East route selection bit and the West route selection bit.

The switch unit 7 transfers the packet to the RPR card to be transmitted according to the information. Here, the switch unit 7 transmits the packet with information indicating whether the packet is destined for East or West.
On the other hand, the RPR card 1 that has received the packet confirms that the packet is addressed to the own card, and then transmits the packet to the ring.

As a result, the packet arrives at the target node via the ring having the shortest path.
In the search of the learning table 10, when a desired transmission destination address is not held in the learning table 10, the packet is subjected to flooding processing as a packet with an unknown transmission destination. This flooding means that when a switch device such as a router receives a packet having an unknown destination address, the packet is sent to all output ports provided in the switch device itself except for the port where the packet is input. Means to copy and send.

  Further, the line card 8 transmits a packet as a broadcast packet to both the RPR cards 1a and 1b at the time of flooding. Then, one of the received RPR cards 1a and 1b transfers the broadcast packet to the ring. On the other hand, when both RPR cards transfer a broadcast packet to the ring, only one of the RPR cards receives the broadcast packet in order to prevent the receiving node from receiving the packet twice. It has become. Of the two types of RPR cards, the RPR card to be transferred is selected based on the setting of the internal packet header.

(8) Configuration of Node Provided with Integrated Card FIG. 8 is a block diagram of a node provided with the integrated card according to the first embodiment of the present invention. The integrated RPR card 1c of node #A shown in FIG. 8 is different from the separate RPR card 1 shown in FIG. 3 in that the external mate interface 6 of the integrated card 1c is connected to another RPR card. There is no point, and the points other than this point are the same. Here, what has the same code | symbol as what was mentioned above has the same thing or those which have the same function.

  Node #A has m ports (not shown) on the outer and inner sides of the quad ring. The integrated card 1c is configured as an integrated card connected to all of the m ports. In other words, the node #A includes an RPR card 1c provided with two RPR units 3a and 3b and a packet processing unit 4, and the RPR card 1c is provided on the outer and inner sides of the multiple packet ring, respectively. It is configured as an integrated card connected to all m ports. In addition, since the nodes #B to #F have the same configuration as the node #A, redundant description is omitted.

Further, the integrated RPR card 1c and the separated RPR cards 1a and 1b share a coupling part for removably inserting into a device rack such as a backboard (not shown). It is also. As a result, a large number of parts can be used in general for card manufacture and card installation, and the cost of the card can be reduced.
Therefore, the integrated card 1c of the present invention switches a packet ring ring packet in which, for example, six nodes #A to #F are connected via a multiple packet ring and a plurality of external network 107 IP packets. The RPR card is configured to send a ring packet and an IP packet to a plurality of external networks 107 or multiple packet rings based on node IDs of received ring packets and destination addresses of IP packets received from a plurality of external networks 107, respectively. Two RPR units 3a and 3b that are input to and output from each other and connected to each other, a ring packet that is output from the two RPR units 3a and 3b to the plurality of external networks 107, and a plurality of external networks 107 IP packets output to the two RPR units 3a and 3b from Each transfer path and a packet processing unit 4 to switch. The integrated card is configured as an integrated card connected to all m ports provided on the outer and inner sides of the multiple packet ring.

As described above, the node #A includes one RPR card 1c instead of the two RPR cards 1a and 1b, and is formed by integrating the functions of the two separate RPR cards 1a and 1b. ing. Therefore, the RPR card 1c can provide both functions of an integrated type and a separated type.
In addition, as described above, the present invention adopts an architecture that realizes the same packet transfer function, so that the switch unit 7 and the line card 8 can be connected regardless of whether a separate card or an integrated card is used. Can be shared. Thereby, cost reduction can be achieved.

(9) Modification (A1) First Modification Each node #A to #F can have various configurations by combining a separation-type card and an integrated card. This will be described with reference to FIGS. In addition, what has the same code | symbol as what was mentioned above in FIGS. 9-11 has the same thing or those same function as them.

FIG. 9 is a block diagram of the node #A according to the first modification of the first embodiment of the present invention. Node #A shown in FIG. 9 is provided with two integrated cards 1c shown in FIG. 8, and is used for quad ring. Therefore, redundancy can be achieved by using the integrated card 1c.
FIG. 10 is a block diagram of an RPR card according to a first modification of the first embodiment of the present invention. The circuit board having the function as the RPR card 1d shown in FIG. 10 is provided with only parts corresponding to one port of the two RPR cards 1a and 1b shown in FIG. Parts corresponding to the RPR portion 3b are not provided. Only the PHY unit 2a and the RPR unit 3a of either East or West (for example, East side) are operated. Further, in node #A, one of the external mate interfaces 6 that connect the two RPR cards 1c is not used, or one of the external mate interfaces 6 is not originally provided. In addition, the node #A is not connected to another RPR card by substituting something other than the external mate interface 6. The node #A shown in FIG. 9 is different in that an external mate interface 6 is provided between the two RPR cards 1a and 1b provided in the node #A shown in FIG.

As a result, the RPR card 1d shown in FIG. 10 has a reduced number of parts and can be made inexpensive.
Next, FIG. 11 shows a node configured using the RPR card 1d.
FIG. 11 is a block diagram of the node #A according to the first modification of the first embodiment of the present invention. The node #A shown in FIG. 11 is configured by providing two separation-type cards 1d shown in FIG. Since this separable card 1d is not connected to other RPR cards, ring redundancy is not necessary.

Therefore, when ring redundancy is not performed, a dual ring having a cost reduction effect can be configured by providing the separation type card 1d.
(A2) Second Modification Two types of interface cards, an integrated card and a separate card, can be mixed in the same node or optical network system, and the integrated card and the separate card are mixed in the same ring. It can also be made.

  Specifically, two RPR cards 1 provided with two RPR units 3a and 3b and a packet processing unit 4 are connected to m ports provided on the outer side and the inner side of the multiple packet ring, respectively. As a separate card in which the first RPR card connected to the outer port and the second RPR card connected to the inner port of the m ports are provided separately, or in the multiple packet ring It can also be configured as an integrated card connected to all m ports provided respectively on the outer and inner sides.

In this case, the packet processing unit 4 is configured to select a failure recovery procedure of m ₁ (m ₁ represents a natural number) including a center wrap and an edge wrap according to an integrated card or a separate card. can do.
Furthermore, the optical network system of the present invention is configured by mixing an integrated card and a separate card in the same packet ring among multiple packet rings.

In this way, communication interruption at the time of exchanging the RPR card is avoided, and continuous communication becomes possible. Further, the cost of system construction can be reduced by reducing the number of RPR cards, etc., and it is possible to make the system reliable even with any configuration of RPR card redundancy and ring redundancy.
In this way, the RPR redundancy method enables RPR card redundancy, shortest path selection, and optimum path selection even when the same RPR card is a separate type and an integrated type. Both route selection and optimum route selection are possible.

(B) Description of Second Embodiment of the Invention Each node #A to #F of the second embodiment dynamically changes the packet transfer path (East side or West side). The optical network system of the second embodiment is the same as the optical transmission system 100 shown in FIG.
FIG. 12 is a configuration diagram of a packet ring according to the second embodiment of the present invention. The packet ring shown in FIG. 12 is provided with, for example, six even-numbered nodes #A to #F. Each node #A to #F in the first embodiment exclusively selects one of the east route selection and the west route selection of the ring topology table 11 (see FIG. 7). In this case, for example, the node #A fixes the selection path for the node #D corresponding to the number of hops 3, for example, and the load on only one of the selected East side or West side becomes heavy. For example, when the node #D is a center station, only the side on which a processing load for a large number of packets is fixedly selected increases.

  For this reason, some or all of the nodes #A to #F in the second embodiment are configured to distribute the load of packet transfer processing using a hash operation in order to select a transfer path. For example, when the node #A has a hash calculation function, the node #A determines the number of hops from the east side and the west side based on the result of the hash calculation on the basis of the ring packet transmission source node (for example, the node #A). The packet is transferred to a node (for example, node #D) having the same number of hops from.

  FIG. 13 is a block diagram of a hash calculation unit according to the second embodiment of the present invention. The hash calculation unit 50 shown in FIG. 13 includes a generation polynomial holding unit 50a that holds a plurality of generation polynomials and hash calculation methods used for hash calculation, a monitoring unit 50b that monitors a load related to ring packet transfer processing, and a monitoring unit A selection unit 50c for selecting a generator polynomial and a hash calculation method held in the generator polynomial holding unit 50a is configured so that the load monitored in 50b is distributed. The broadcast register 50f will be described later with reference to FIG.

  The selection unit 50c sets the node #A and the destination node (for example, node #D) for each of the outer side and the inner side of the multiple packet ring such as the dual ring or the quad ring starting from the node #A of the source node. A generator polynomial is selected by hash operation for a node #D having the same number of hops of relay nodes of ring packets between them. Here, the relay nodes correspond to nodes #B and #C or nodes #E and #F, and the number of hops is three.

Since there are m ₂ (m ₂ represents a natural number) types of generator polynomials used for the hash operation, each node #A to #F has, for example, a packet processing unit 4 or a generator polynomial holding unit 50a specialized for the hash operation. Provided, and m ₁ types of generator polynomials used for the hash calculation are registered in advance in the generator polynomial holding unit 50a. And each node # A- # F monitors the packet amount transmitted to a ring as statistical information, and when the big bias has occurred in the load distribution result (and it is likely to occur), The generator polynomial used for the hash operation is changed to another generator polynomial, and dynamically changed to values necessary for the generator polynomial and the hash operation.

  The ring select table (hash calculation result holding unit) 50d will be described later with reference to FIG. 13 is provided in the packet processing unit 4, but may be provided separately from the packet processing unit 4 in the RPR card 1. The RPR card 1 in the second embodiment is the same as or similar to the RPR card 1 described above, and RPR cards 1a, 1b, 1c, and 1d can be used.

  In the following description, the number of nodes is 6. However, even when the number of nodes is an even number such as 8, 10, the load distribution processing for packet transmission similar to the case of the number of nodes 6 is performed. Note that the ring topology table 11a may be provided in some of the nodes #A to #F. Also in the second embodiment, the RPR card 1 has one function or both functions of the integrated type and the separated type.

FIG. 14 is a diagram showing an example of a ring topology table according to the second embodiment of the present invention. Each node #A to #F in the second embodiment is added to the ring topology table 11a shown in FIG. Whether or not to perform a hash operation is selected with reference to the item “hash operation”.
In addition, each item of East route selection, West route selection, and hash calculation is set exclusively. Specifically, when East route selection is off, West route selection is off, and hash computation is on, node #A performs hash computation on the destination address and source address of the ring packet, Either East or West is selected based on the result of the hash calculation.

  Further, the information required for each node #A to #F for the hash calculation is not fixed to one of various field information included in the ring packet, but desired information is used. This is because each node #A to #F performs calculations using information suitable for hash calculation according to the specifications of the optical network system 100 and various sub-network systems. The information to be calculated is, for example, a MAC address or an IP address, and any of these can be selected.

As a result, each of the nodes #A to #F avoids an excessive load in only one direction when transmitting a packet to nodes having the same number of hops, and can distribute the load. Further, each of the nodes #A to #F can dynamically control load distribution in the ring network.
The above description is a method of selecting a transfer path for a normal packet using the calculation result of the hash calculation.

On the other hand, the calculation result of the hash calculation can also be used for the selection of the route of the broadcast packet and the selection of the transfer ring of the broadcast packet.
Each node #A to #F can also distribute the load using the hash calculation for the distribution of the broadcast packet. Also, the hash operation does not depend on the multiplex number. Hereinafter, a hash calculation in the case where a triple packet ring is used will be described.

FIG. 15A is a configuration diagram of a triple packet ring according to the second embodiment of the present invention. The triple packet ring shown in FIG. 15A has rings # 1, # 2, and # 3.
Here, the ring select table 50d provided in the hash calculation unit 50 (see FIG. 13) holds correspondence data that associates the calculation result of the hash calculation with the ring ID, and serves as a hash calculation result holding unit. It is functioning.

  FIG. 15B is a view showing an example of the ring select table according to the second embodiment of the present invention. The ring select table 50d shown in FIG. 15B is an operation result obtained by a hash operation. “0” representing (remainder) and the ring ID “ring # 1” are held in association with each other. Then, the packet processing unit 4 provided in each of the nodes #A to #F updates the corresponding data held in the ring select table 50d when a failure occurs in the multiple packet ring, so that the packet ring in which the failure has occurred is updated. Broadcast packet distribution via the network is stopped, and load distribution of broadcast packet distribution is continued.

In other words, the nodes #A to #F according to the present invention, when there are an even number of nodes #A to #F, such as six, are present in the packet ring, the calculation result of the hash operation is the route selection of the broadcast packet. To be used.
In addition, in the nodes #A to #F of the present invention, for example, when a triple packet ring such as a triple is configured, the calculation result of the hash calculation is used to select the transfer ring of the broadcast packet.

The nodes #A to #F according to the present invention are such that, for example, when an even number of nodes #A to #F, such as six, exist in the packet ring, the calculation result of the hash calculation is used for route selection of the broadcast packet. It has become.
In the hash calculation method in load distribution, the value of the address field included in the packet header is divided using a generator polynomial, and the route distribution of the packet is performed using the remainder value obtained by the calculation. Here, in the case of a triple packet ring, the packet processing unit 4 divides the address field using a generator polynomial having a maximum remainder value “2”, and sets the remainder value and the ring ID in the ring select table 50d. Link to 1 and hold. For example, when the remainder value of the hash calculation result of the broadcast packet is “1”, ring # 2 is used as the broadcast packet transfer destination ring.

Further, the packet processing unit 4 dynamically updates the generator polynomial and the hash calculation method held in the generator polynomial holding unit 50a, thereby stopping broadcast packet distribution through the failed packet ring and broadcasting. Continue to distribute packet distribution load.
Thus, each node #A to #F dynamically updates the hash calculation result and the ring ID held in association with the ring select table 50d when a failure occurs in the packet ring.

Next, processing when a failure occurs at a plurality of locations will be described with reference to FIGS. 16 (a) and 16 (b).
FIG. 16A is a diagram for explaining a triple packet ring when a failure occurs according to the second embodiment of the present invention. This is a case where a failure occurs in ring # 2 between node #C and node #D and between node #D and node #E shown in FIG.

  Here, if there is a failure only between the node #C and the node #D, each node #A to #F may update the ring topology table 11a in order to avoid the failure part in the ring # 2. On the other hand, when a failure occurs between the node #D and the node #E and a failure occurs in combination, the node #D avoids packet transmission to the ring # 2, and broadcast packets to the rings # 1 and # 3. It is necessary to cause the other nodes to learn a new ring ID by distributing.

FIG. 16B shows an example of a ring select table according to the second embodiment of the present invention. Each node #A to #F has linked the ring # 1 to the ring select table 50d shown in FIG. 16B to avoid transmission to the ring # 2 with the remainder of the calculation result being “1”. Write data.
Thus, using the hash calculation result, the node does not transmit a broadcast packet for ring # 2, and transmits a packet for ring # 1 and ring # 3.

As described above, when a failure occurs in the packet ring, each of the nodes #A to #F dynamically changes the generator polynomial and the hash calculation method used for the hash calculation, and updates the link select table 50d. As a result, broadcast packet distribution to the failure ring is stopped, and load distribution of broadcast packet distribution continues.
Also, other patterns different from the above-described patterns will be described with reference to FIGS. 17 (a), 17 (b ) to 18 (a), and 18 (b).

FIG. 17A is a diagram showing an example of another triple packet ring according to the second embodiment of the present invention. The triple packet rings shown in FIG. 17A are connected using rings # 1 to # 3.
FIG. 17B is a diagram showing an example of another ring select table according to the second embodiment of the present invention. The link select table 50d shown in FIG. 17 (b) holds the calculation result obtained by the hash calculation and the ring ID linked to each other. As a result, the same processing as in the packet ring shown in FIGS. 15A and 15B is performed.

Next, a case where a failure occurs at two locations in the packet ring will be described.
FIG. 18A is a diagram showing an example of a triple packet ring when a failure occurs according to the second embodiment of the present invention. This is a case where a failure occurs in the ring # 2 between the node #C and the node #D and between the node #D and the node #E shown in FIG. Each node #A to #F changes the generator polynomial itself when a failure occurs.

  Here, when a failure occurs, the nodes #A to #F can update the ring select table 50d itself without changing the generator polynomial in order to reduce the set amount. For example, FIG. In the ring select table 50d shown in FIG. 3), the bias toward the ring # 1 becomes large and the load may be concentrated. Therefore, in order to distribute the load, when a failure occurs, the nodes #A to #F dynamically change the generator polynomial itself so that “the remainder value of the operation result = all nodes−failure node”. It is.

FIG. 18B is a diagram showing an example of the ring select table 50d according to the second embodiment of the present invention. When the data in the ring select table 50d shown in FIG. 18B is updated and a failure occurs. In this case, optimal load distribution can be achieved.
As described above, according to the node or RPR packet transfer method shown in FIGS. 16A and 16B, the generator polynomial is not changed and the ring select table 50d itself is updated. The amount is small.

Further, according to the nodes #A to #F shown in FIGS. 18A and 18B, when a failure occurs, the results of packet distribution (load distribution) are equalized.
(B1) First Modification FIG. 19 is a configuration diagram of a packet ring according to a first modification of the second embodiment of the present invention. The six nodes #A to #F shown in FIG. 19 control the TTL value of the broadcast packet, and transmit the broadcast packet to both the East and West packet rings. The load is distributed.

  In the first embodiment, the broadcast packet is transmitted to either the East or West ring. Then, each node #A to #F of the first embodiment uses a characteristic of a ring type to transmit a packet in any one direction or any one ring, and the broadcast packet circulates around the packet ring. Each node #A to #F receives the broadcast packet, and finally the source node deletes the broadcast packet from the ring.

In this case, the load of the broadcast packet in a specific ring becomes heavy.
For this reason, each of the nodes #A to #F in this modification is provided with a broadcast register 50f shown in FIG. The broadcast register 50f can hold the following items, and this function is exhibited by a register, for example. Note that the selection unit 50c shown in FIG. 13 is provided with a read / write processing unit (not shown) capable of reading and writing the broadcast register 50f.

The items of the broadcast register are East route selection, West route selection, EastTTL, WestTTL, and hash calculation. Each node #A to #F refers to the contents of the broadcast register 50f every time a broadcast packet is transmitted, and transfers the packet according to the set contents of the broadcast register 50f.
FIG. 20A is a diagram showing items of the broadcast register according to the first modification of the second embodiment of the present invention. The broadcast register 50f shown in FIG. 20A is for transferring a broadcast packet from the node #A, and the East route selection is off, the West route selection is off, and the hash calculation is off. . In this case, the EastTTL value and the WestTTL value are recognized as valid, and the broadcast packet is transferred based on these TTL values (EastTTL is 3 and WestTTL is 2). Specifically, the broadcast packet is transmitted from the packet ring to the external network 107 via the node #A, the node #B, the node #C, the node #D, the node #A, the node #F, and the node #E, respectively. Is done.

  FIG. 20B is a diagram showing a setting example of the broadcast register 50f according to the first modification of the second embodiment of the present invention. Packets are transmitted based on the setting contents shown in Example 4 from Example 1 (same as FIG. 20A) shown in FIG. Each item of East route selection, West route selection, and hash calculation is set exclusively, and the TTL setting is made valid when all items are off.

  Furthermore, both the East route selection and the West route selection are used when a ring for transmitting a broadcast packet is fixed. For example, when the number of nodes is 6, the EastTTL value and the WestTTL value are set to 3 and 2, respectively, and a broadcast packet by selecting the shortest route becomes possible. That is, the packet processing unit 4 selects a dedicated transmission path for transmitting a broadcast packet from among the multiple packet rings when the multiple packet ring is quadruple or more.

In this case as well, when the number of nodes existing in the ring is an even number, in order to prevent the load on the nodes having the same number of hops from being counted from either the East or West direction, the hash calculation is performed. When is set to ON, load distribution of broadcast transfer by hash calculation can be performed.
The broadcast register 50f can be provided in either the line card 8 or the RPR card 1 (or the RPR cards 1a to 1d). When provided in the line card 8, the separation-type card needs to transmit a broadcast packet together with TTL information to both the East side card and the West side card.

  On the other hand, when the broadcast register 50f is provided in the RPR card 1, in the separation type card, in the line card 8, the broadcast packet is transferred to both the East side card and the West side card, and the East side card and the West side card are connected. Each refers to the broadcast register 50f and transmits a broadcast packet to the packet ring in accordance with the setting contents.

When the RPR card 1 of each of the nodes #A to #F does not need to transfer a broadcast packet, the RPR card 1 discards the broadcast packet. For example, when East route selection is on and West route selection is off, the West side RPR card 1 (or RPR cards 1a to 1d) discards.
As described above, in the nodes #A to #F of the present invention, when an even number of nodes #A to #F exist in the packet ring, the calculation result of the hash calculation is used for route selection of the broadcast packet.

In addition, in the nodes #A to #F of the present invention, for example, when a multiple packet ring such as triple is configured, the calculation result of the hash calculation is used for selection of the transfer ring of the broadcast packet.
As described above, by using the hash operation, the same effect as that of the first embodiment can be obtained, and the load of each of the nodes #A to #F can be distributed.

(B2) Second Modification The second modification relates to load distribution of broadcast transfer processing in a quad ring configuration. In this modification, when the packet ring is equal to or higher than the quad ring, each node #A to #F constituting the packet ring is given a function of selecting a packet ring for transmitting a broadcast packet. The load of broadcast packet transmission processing of each node #A to #F is distributed.

On the other hand, when a packet ring for transmitting a broadcast packet is fixedly set for each node, when attention is paid to a specific node, the load of the broadcast packet on the specific ring becomes excessive.
For this reason, the packet processing unit 4 of each of the nodes #A to #F in the second modified example has a ring selection function that does not depend on the shortest path of the ring packet (giving the ring selection function), so The optimum route of the ring packet is selected according to the bandwidth used for the multiple packet ring.

Here, a quad ring (quadruple ring) can be configured by a method in which a node uses a separate card and a method in which a node uses two integrated cards.
FIG. 21 is a configuration diagram of a quad ring according to a second modification of the second embodiment of the present invention, in which nodes #A to #F are all connected to ring # 1 and ring # 2, and the ring Are connected to both rings # 1 and # 2. The case where all the nodes #A to #F are not connected to the rings # 1 and # 2 will be described later.

  FIG. 22 is a diagram showing an example of a broadcast register according to a second modification of the second embodiment of the present invention. The broadcast register 60 shown in FIG. 22 is a “ring” representing a ring number for transmitting a broadcast packet. The item “selection” is provided, and each node #A to #F refers to this “ring selection” to select a packet ring for transmitting a broadcast packet. Note that the items such as “East route selection” in the broadcast register 60 shown in FIG. 22 are the same as the items shown in FIGS. 20 (a) and 20 (b).

FIG. 23 is a diagram showing a second example of the broadcast register according to the second modification of the second embodiment of the present invention. The broadcast register 61 shown in FIG. 23 performs “ring selection” and “route selection” using a hash operation.
With this configuration, for example, when node #A shown in FIG. 21 transmits a broadcast packet to both ring # 1 and ring # 2, each node #A to #F receives the broadcast packet twice. Each of the nodes #A to #F monitors and manages, for example, by assigning a priority class to each of the rings # 1 and # 2 using the rings # 1 and # 2 as primary rings and secondary rings, respectively. That is, each node #A to #F validates the broadcast packet received from the primary ring and discards the broadcast packet received from the secondary ring. The priority class is assigned according to the number of rings when the number of rings is three or more.

  As a result, each node #A to #F selects a ring for transmitting a broadcast packet, and each node #A to #F has “East route selection” and “West route” set in the broadcast registers 60 and 61. Based on each item of “selection”, “hash operation”, EastTTL ”and“ WestTTL ”, an optimum route of the broadcast packet is selected in the selected ring.

Therefore, the load of the broadcast packet transmission processing of each node #A to #F is distributed.
In this method, since the broadcast packet from the node is transmitted to only one of the rings, the receiving node can process the received packet as it is without being aware of the primary ring and the secondary ring, and the load is increased. Distributed. Furthermore, since it is possible to set a ring in which a broadcast packet should be transmitted in units of each node #A to #F, the node #A can use the ring # 1 and the other node #B can use the ring # 2, The load distribution of broadcast packet transmission in quad ring becomes possible.

(B3) Third Modification The third modification relates to a learning table in a ring redundant configuration with a quad ring or more and in a ring redundant configuration in which the ring topology data state of each ring is different. In this case, each node #A to #F checks with reference to the source node ID of the broadcast packet and the ring topology table 11a, and receives only packets transferred via the shortest path set in the ring topology. However, other broadcast packets are discarded. As a result, the shortest path learning for broadcast packet transfer is possible.

  In other words, the learning table in the third modified example is transferred via the shortest path based on the node ID of the source node of the broadcast packet and the ring topology when the multiple packet ring is quadruple or more. The shortest path for transferring the broadcast packet is learned by acquiring the ring packet and discarding the broadcast packet.

  FIG. 24 is a configuration diagram of a quad ring according to a third modification of the second embodiment of the present invention. Nodes #A, #C, #D, and #F shown in FIG. 24 are connected to both rings # 1 and # 2, node #B is connected only to ring # 1, and node #E is connected to ring # 1. Only connected to 2. For this reason, the ring topologies of ring # 1 and ring # 2 are different. Therefore, a process in which each node #A to #F only transmits a broadcast packet to either ring # 1 or ring # 2 is performed by a node that cannot receive a broadcast packet (of nodes #A to #F). Cause any node) to occur.

For this reason, each node #A to #F in the third modified example transmits a broadcast packet to both rings # 1 and # 2, and each node connected to both rings # 1 and # 2 is autonomous. Broadcast packets are received only from the selected packet ring, and broadcast packets received from rings other than the selected ring are discarded.
FIG. 25 is a diagram showing an example of a ring topology table according to the third modification of the second embodiment of the present invention. The ring topology table 11b shown in FIG. 25 is created for the ring # 1 and the ring # 2 in the RPR card 1 (or the RPR cards 1a to 1d) in each of the nodes #A to #F. Each node #A to #F transmits a broadcast packet to both rings # 1 and # 2.

With such a configuration, a processing method will be described using the node #A as an example. Since the nodes #B to #F perform the same processing as the node #A, redundant description is omitted.
When the node #A receives the broadcast packet, the packet processing unit 4 searches the ring topology table 11b with reference to the node number of the transmission source ring of the received packet. As a result, when the received packet is a broadcast packet received from the ring having the shortest route, the packet processing unit 4 receives the ring packet and transmits the broadcast packet to the line card 8 (see FIG. 3). . If the received packet is not a broadcast packet, the broadcast packet is discarded.

Here, the node #A sets the ring # 1 as a master ring. That is, the node #A is made different from the nodes #B to #F as slave rings. Then, the node #A receives the broadcast packet from the node #C from the ring # 1, and discards the packet from the ring # 2.
In this case, since the ring number that received the broadcast packet is learned in the learning table 10 of the line card 8 at the node #A, the ring # 1 is learned between the node #A and the node #C. Therefore, the node #A selects the ring # 1 when the packet is next transmitted from the node #A to the node #C. Therefore, it is preferable that the learning table of the node #A learns the ring # 2 that is the shortest route between the node #A and the node #C in consideration of the ring use efficiency.

  Further, when the source node number of the broadcast packet is, for example, the node #C, the node #A searches the ring topology table 11b to receive the shortest path (hop count is 1) from the east port of the ring # 2. Recognize that. Therefore, when this broadcast packet is received from the East port of ring # 2, node #A captures the broadcast packet, transmits the captured packet to the line card 8, and other than the East port of ring # 2. Drop packets from.

Thus, the learning table 10 of the line card 8 can learn about the packet ring that is the shortest path, and can be transmitted efficiently.
(B4) Fourth Modification The optical network system in the fourth modification is obtained by adding an optimum path selection function to the shortest path selection function of the third modification. That is, the learning of the shortest path of the optical network system is performed in each node #A in a ring redundant configuration in which the ring topology data state of each packet ring is different in a ring redundant configuration of quadring or higher, as in the third modification. ~ # F is checked by referring to the source node ID of the broadcast packet and the ring topology table 11a, receives only the ring packet transferred through the shortest path set in the ring topology, and broadcasts other than that Discard the packet.

In this modification, each of the nodes #A to #F is provided with a ring selection function that does not depend on the shortest route, and thereby selects the optimum route according to the shortest route selection and the bandwidth usage state of the network.
In the quad ring shown in FIG. 24, when each line capacity of each ring is, for example, 1 Gbps, each node #A to #F is assigned a packet transfer amount between each node #A to #F itself. Specifically, the packet transfer amount from node #A to node #B is set to 200 Mbps, the packet transfer amount from node #A to node #C is set to 600 Mbps, and packet transfer from node #A to node #D is further transferred. The amount is 600 Mbps.

Here, the packet addressed to the node #C from the node #A and the packet addressed to the node #D from the node #A respectively use the east of the ring # 2, which is the shortest path, and thus exceed the line capacity of 1 Gbps. Therefore, for example, a packet addressed to node #C from node #A preferably uses East of ring # 1.
In the packet ring, the dynamic switching of the packet transmission ring depending on the line, the traffic amount of the system, the used bandwidth, or the like causes the packet order to be reversed before and after the path switching. Therefore, it is preferable that each node #A to #F statically set the ring topology table 11a shown in FIG.

FIG. 26 is a diagram showing an example of a ring topology table according to the fourth modification of the second embodiment of the present invention. The ring topology table 11c shown in FIG. 26 is provided with a ring forced selection bit, so that an optimum packet ring can be selected regardless of the shortest path.
For example, when node #A receives a broadcast packet from node #C, node #A refers to ring topology table 11c. Here, regarding the number of hops, ring # 2 is the shortest path, but node #A selects ring # 1 in which the ring forcible selection bit is on, and the ring forcible selection bit For example, for node #D in which is off, node #A selects the packet ring that provides the shortest path.

Thereby, it is possible to select both the shortest path and the optimum path in the quad ring configuration.
As described above, according to the RPR cards 1a to 1d according to the first to fourth modifications of the second embodiment and the second embodiment, each of the nodes #A to #F uses the same RPR card. In addition to realizing both the body type and the separation type, the packet ring of the separation type card and the integral type card can be realized by using common hardware.

As described above, a common processing method can be used for packet processing in each of the nodes #A to #F, and the line card 8 and the switch unit 7 (see, for example, FIG. 3) can be shared.
The packet ring can be made redundant suitable for the connection topology of the nodes #A to #F, regardless of the card redundancy or ring redundancy method, and contributes to improving the reliability of the packet ring.

In addition to these, in the packet ring, each of the nodes #A to #F can perform card redundancy, shortest path selection, and optimum path selection when either an integrated card or a separate card is provided. Further, the shortest path selection and the optimum path selection are possible even in a quadruple or more packet ring.
(C) Description of Third Embodiment of the Invention In the third embodiment, a protection method will be described for each of the integrated and separated node #A to #F configurations. When a failure occurs in a packet ring node or ring (transmission path), the node closest to the location where the failure has occurred usually sends a failure occurrence message to the ring, and each of the nodes #A to #F has wrap protection ( Start Wrap Protection) operation. By switching the packet ring of each RPR card (the above-described RPR cards 1a and 1b unless otherwise specified), the ring forms a closed loop.

  In the third embodiment, each node #A to #F selects the wrap protection function between the center wrap and the edge wrap according to the configuration of each of the integrated and separated nodes #A to #F. It has become. The center wrap is protection for returning a packet between two RPR cards 1a and 1b, and the edge wrap is protection for returning a packet at a portion where the RPR cards 1a and 1b are connected to a ring. That is, the difference between the center wrap and the edge wrap is different in the location where the lap is activated.

  FIG. 27 is a view for explaining a center wrap according to the third embodiment of the present invention. When one of the two RPR cards 1a and 1b provided in the node #A shown in FIG. 27 stops operation due to failure or maintenance, the node #A Activate wrap protection. For example, when a failure occurs in the East side RPR card 1a, the node #A starts the wrap protection by the West side RPR card 1b and starts center wrapping. In addition, since the node #B activates the wrap, a closed loop using wrap protection can be formed even when the RPR card 1a (East) of the node #A is removed from the system by maintenance replacement.

FIG. 28 is a view for explaining edge wrap according to the third embodiment of the present invention. When a line failure occurs between the node #A and the node #B shown in FIG. 28, a closed loop can be formed by the node #A starting the edge wrap and the node #B starting the wrap protection. It becomes.
Therefore, it has lap protection functions for both center wrap and edge wrap, and each node #A to #F selects the wrap protection function for center wrap and edge wrap according to each of the integrated and separated types. it can.

Conventionally, a node provided with an integrated card supports only edge wrap. This is because the integrated card is removed together with the RPR card (see FIGS. 33 and 35) at the time of maintenance and replacement, and therefore, protection cannot be performed even if the center wrap is mounted.
On the other hand, the nodes #A to #F according to the present invention have both the center wrap and the edge wrap on the RPR cards 1c and 1d, regardless of whether the RPR card is an integrated card or a separate card. By implementing the wrap protection, a desired wrap protection can be achieved according to each configuration of the integral type or the separated type.

(D) Others The present invention is not limited to the above-described embodiments and modifications thereof, and various modifications can be made without departing from the spirit of the present invention.
The number of nodes constituting the packet ring can be variously changed, such as 2, 4, 6, 8 and 8 or more.

  As described above in detail, according to the node of the present invention, the optical network system of the integrated card and the separate card can be realized by using common hardware, and the packet processing in each of the nodes #A to #F is common. The processing method can be used, and the line card and the switch card can be shared, thereby reducing the cost of system construction.

In the case of the integrated card, the node can be connected to the packet ring using one RPR card, and the cost of the packet ring and the node can be reduced as compared with the case of the separation type card. In addition, since it is possible to reduce the consumption of slots for accommodating RPR cards, it is possible to increase the number of ports of the entire node including the line cards.
In particular, it is prominent when used in a small-scale network or the like in which cost reduction is more important than reliability. Accordingly, the service is improved with respect to a carrier network and the like owned by a communication carrier (communication carrier) who places importance on reliability rather than cost reduction.

  1 is a configuration diagram of an optical network system according to a first embodiment of the present invention.   1 is a schematic configuration diagram of an RPR network according to a first embodiment of the present invention.   It is a block diagram of a node provided with a separation type card according to the first embodiment of the present invention.   1 is a block diagram of an RPR interface card according to a first embodiment of the present invention.   It is a figure for demonstrating the switch process in the separation-type card structure which concerns on 1st Embodiment of this invention.   (A)-(c) is a figure which shows the packet format which concerns on 1st Embodiment of this invention.   It is a figure for demonstrating the ring topology table which concerns on 1st Embodiment of this invention.   It is a block diagram of a node provided with an integrated card according to the first embodiment of the present invention.   It is a block diagram of a node concerning the 1st modification of a 1st embodiment of the present invention.   It is a block diagram of the RPR interface card which concerns on the 1st modification of 1st Embodiment of this invention.   It is a block diagram of a node concerning the 1st modification of a 1st embodiment of the present invention.   It is a block diagram of the packet ring which concerns on 2nd Embodiment of this invention.   It is a block diagram of the hash operation part which concerns on 2nd Embodiment of this invention.   It is a figure which shows an example of the ring topology table which concerns on 2nd Embodiment of this invention.   (A) is a block diagram of the triple packet ring which concerns on 2nd Embodiment of this invention, (b) is a figure which shows an example of the ring selection table which concerns on 2nd Embodiment of this invention.   (A) is a figure for demonstrating the triple packet ring at the time of the failure generation which concerns on 2nd Embodiment of this invention, (b) shows an example of the ring selection table which concerns on 2nd Embodiment of this invention. FIG.   (A) is a figure which shows an example of the other triple packet ring which concerns on 2nd Embodiment of this invention, (b) is a figure which shows an example of the other ring select table which concerns on 2nd Embodiment of this invention. It is.   (A) is a figure which shows an example of the triple packet ring at the time of the failure generation which concerns on 2nd Embodiment of this invention, (b) is a figure which shows an example of the ring select table which concerns on 2nd Embodiment of this invention. It is.   It is a block diagram of the packet ring which concerns on the 1st modification of 2nd Embodiment of this invention.   (A) is a figure which shows the item of the broadcast register which concerns on the 1st modification of 2nd Embodiment of this invention, (b) is the setting of the broadcast register which concerns on the 1st modification of 2nd Embodiment of this invention. It is a figure which shows an example.   It is a block diagram of the quad ring which concerns on the 2nd modification of 2nd Embodiment of this invention.   It is a figure which shows an example of the broadcast register which concerns on the 2nd modification of 2nd Embodiment of this invention.   It is a figure which shows the 2nd example of the broadcast register which concerns on the 2nd modification of 2nd Embodiment of this invention.   It is a block diagram of the quad ring which concerns on the 3rd modification of 2nd Embodiment of this invention.   It is a figure which shows an example of the ring topology table which concerns on the 3rd modification of 2nd Embodiment of this invention.   It is a figure which shows an example of the ring topology table which concerns on the 4th modification of 2nd Embodiment of this invention.   It is a figure for demonstrating the center wrap which concerns on 3rd Embodiment of this invention.   It is a figure for demonstrating the edge wrap which concerns on 3rd Embodiment of this invention.   It is a figure which shows the structural example of an optical network system.   (A), (b) is a figure which shows an example of the packet format used for an RPR network, respectively.   (A)-(c) is a figure for demonstrating a multiple ring transmission path, respectively.   It is a block diagram of the separation type | mold RPR interface card provided in the node of the RPR network.   It is a block diagram of the integrated RPR interface card provided in the node of the RPR network.   It is a figure which shows the data path in the conventional separation type | mold RPR interface card.   It is a figure which shows the data path in the conventional integrated RPR interface card.

Claims (10)

A node that switches a ring packet of a ring transmission path in which a plurality of nodes are connected via a multiple ring transmission path and a second packet of a plurality of second transmission paths;
A node identifier representing a transmission source node of the ring packet, a ring identifier representing each ring transmission path of the multiple ring transmission path, address information of a transmission source device of the second packet, and connection states of the plurality of nodes. A line card that transmits and receives the second packet to and from the second transmission line based on the ring topology that represents
Based on the node identifier of the received ring packet and the destination address of the second packet received by the line card, the ring packet and the second packet are respectively transmitted to the line card side or the multiple ring transmission line side. A plurality of protocol processing units input / output between and connected to each other;
A packet processing unit for switching each transfer path between the ring packet output from the plurality of protocol processing units to the line card and the second packet output from the line card to the plurality of protocol processing units; A node, characterized by comprising A plurality of RPR interface cards provided with the plurality of protocol processing units and the packet processing unit;
At least one of the plurality of RPR interface cards is
A first card connected to a port in the first direction among a plurality of ports provided in each of the first direction and the second direction of the multiple ring transmission line, and a second direction of the plurality of ports 2. The node according to claim 1 , wherein the node is configured as a separate card provided separately from the second card connected to the port. An RPR interface card provided with the plurality of protocol processing units and the packet processing unit;
The above RPR interface card is
2. The node according to claim 1 , wherein the node is configured as an integrated card connected to all of a plurality of ports respectively provided in the first direction and the second direction of the multiplex ring transmission line. The plurality of protocol processing units respectively include a first protocol processing unit that inputs and outputs the ring packet with respect to a part of the ring transmission lines of the multiple ring transmission lines, A second protocol processing unit for inputting and outputting the ring packet to and from the remaining ring transmission line,
The first protocol processing unit and the second protocol processing unit provided in the same RPR interface card among the plurality of RPR interface cards are connected via an internal mate interface, and the plurality of RPR interfaces characterized in that the first protocol processing unit provided in a separate RPR interface cards of the card and the second protocol processing unit is configured by connecting through the external mate interface, claim 2 or 3. The node described in 3 . The plurality of protocol processing units are respectively
Based on the destination address of the ring packet, it determines whether the ring packet is a to some other node destined addressed to its own node,
If it is determined that the own node addressed, while transferring those the ring packet to the packet processing unit, if it is determined that the other nodes addressed and configured to our the ring packet to through to said multiplexing ring transmission path The node according to claim 4, wherein: A hash operation result holding unit for holding correspondence data in which a calculation result of a hash operation and the ring identifier are associated with each other is provided.
When a failure occurs in the multiplex ring transmission line, by updating the corresponding data held in the hash calculation result holding unit, broadcast packet distribution through the ring transmission line in which the failure has occurred is stopped, and the broadcast packet The node according to claim 1 , wherein the node is configured to continue load distribution of distribution. The line card
A learning table that holds the node identifier, the ring identifier, and address information of the transmission source device of the second packet in association with each other; a ring topology table that holds a ring topology that represents a connection state of the plurality of nodes; With
The learning table is
When the multiplex ring transmission line is four or more layers, the ring packet transferred via the shortest path is acquired and the broadcast packet is discarded based on the node identifier of the source node of the broadcast packet and the ring topology. The node according to claim 1 , wherein the node is configured to learn a shortest path for transferring the broadcast packet. An RPR interface card that switches a ring packet of a ring transmission path in which a plurality of nodes are connected via a multiple ring transmission path and a second packet of a plurality of second transmission paths,
Based on the node identifier of the received ring packet and the destination address of the second packet received from the plurality of second transmission paths, the ring packet and the second packet are respectively sent to the plurality of second transmission paths or A plurality of protocol processors connected to each other and input / output to / from the multiple ring transmission line side;
Each transfer path between the ring packet output from the plurality of protocol processing units to the plurality of second transmission paths and the second packet output from the plurality of second transmission paths to the plurality of protocol processing sections. And a packet processing unit for switching
A first card connected to a port in the first direction among a plurality of ports provided in each of the first direction and the second direction of the multiple ring transmission line, and a second direction of the plurality of ports An RPR interface card, characterized in that it is configured as a separate card provided separately from the second card connected to the other port. An RPR interface card that switches a ring packet of a ring transmission path in which a plurality of nodes are connected via a multiple ring transmission path and a second packet of a plurality of second transmission paths,
Based on the node identifier of the received ring packet and the destination address of the second packet received from the plurality of second transmission paths, the ring packet and the second packet are respectively sent to the plurality of second transmission paths or A protocol processing unit that inputs and outputs to and from the multiple ring transmission line side;
A packet for switching each transfer path between the ring packet output from the protocol processing unit to the plurality of second transmission paths and the second packet output from the plurality of second transmission paths to the protocol processing unit. With a processing unit,
An RPR interface card configured as an integrated card connected to all of a plurality of ports respectively provided in the first direction and the second direction of the multiplex ring transmission line. An optical network system provided with a node for switching a ring packet of a ring transmission path in which a plurality of nodes are connected via a multiple ring transmission path and a second packet of a plurality of second transmission paths,
At least one of the plurality of nodes or the ring transmission line is
Based on the node identifier of the received ring packet and the destination address of the second packet received from the plurality of second transmission paths, the ring packet and the second packet are respectively sent to the plurality of second transmission paths or A plurality of protocol processors connected to each other and connected to the multiple ring transmission line side; the ring packets output from the plurality of protocol processors to the second transmission line side; An RPR interface card provided with a packet processing unit for switching each transfer path with the second packet output from the two transmission paths to the plurality of protocol processing units;
A first card connected to a port in the first direction among a plurality of ports provided in each of the first direction and the second direction of the multiple ring transmission line, and a second direction of the plurality of ports A separate card provided separately from the second card connected to the other port, and one connected to all of the plurality of ports respectively provided in the first direction and the second direction of the multiplex ring transmission line. An optical network system, characterized in that it is configured by mixing body cards.

Images