JP6028705B2 - Network relay system and switch device - Google Patents

Description

  The present invention relates to a network relay system and a switch device, and, for example, relates to a maintenance / management technique for a network relay system in which link aggregation is applied to communication from one switch device to a plurality of switch devices.

  For example, Patent Document 1 describes a method in which a source address included in a broadcast frame is registered in an FDB at a plurality of upper switches at the same time. Specifically, the lower switch operates in a state in which LAG is set to a link with a plurality of upper switches, and when receiving a broadcast frame such as an ARP request frame, the lower switch is associated with the LAG. Broadcast to multiple higher-level switches.

  Further, in Patent Document 2, in a network relay system including a plurality of fabric repeaters and a plurality of interface repeaters, a plurality of links connecting each interface repeater and a plurality of fabric repeaters respectively. A method for automatically setting a LAG is shown.

JP 2012-249050 A JP 2013-70297 A

  For example, as shown in Patent Document 1 and Patent Document 2, a technique for constructing a network relay system by combining a plurality of box type switch devices instead of a chassis type switch device is known. The network relay system includes a plurality of box-type switch devices (referred to as port switches here) and a plurality of box-type switch devices (referred to as fabric switches here) for relaying frames between the port switches. It is done. Each port switch has a link with a plurality of fabric switches, and sets a link aggregation group (hereinafter abbreviated as LAG) for the plurality of links. In this specification, such a network relay system is called a box-type fabric system.

  Under such circumstances, there are cases where it is desired to perform maintenance / management such as information collection, various settings, firmware update, and the like for the fabric switch in such a box-type fabric system. In this case, in general, a dedicated port for maintenance / management (management port) is provided for the fabric switch in addition to the normal port connected to the port switch, and the dedicated port is used for maintenance / management. A method is used in which terminals are connected by a dedicated communication line. However, in this case, the number of ports, communication lines, etc. will increase, so maintenance and management will be realized using ordinary ports and communication lines instead of dedicated ports and communication lines (ie, in-band management will be realized) ) Is required.

  On the other hand, as disclosed in Patent Document 1, each port switch may have a function of synchronizing an address table of each fabric switch by broadcasting a broadcast frame toward each fabric switch. In this specification, this function is called an address table synchronization function. As shown in Patent Document 1, the address table synchronization function includes a function for preventing side effects caused by the broadcast processing described above. Specifically, each port switch has a function of receiving only one ARP request frame relayed through each fabric switch.

  However, as a result of studies by the present inventors, when in-band management is performed with such an address table synchronization function installed in each port switch, communication using an ARP request frame cannot be performed during in-band management. It has been found that can occur. In this case, sufficient in-band management cannot be realized, and it may be difficult to improve the maintainability of the system.

  The present invention has been made in view of the above, and one of its purposes is to realize improvement in maintainability in a switch device and a network relay system including the switch device.

  The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

  Of the inventions disclosed in the present application, the outline of a typical embodiment will be briefly described as follows.

  The network relay system according to this embodiment includes n (n is an integer of 2 or more) port switches and m (m is an integer of 2 or more) fabrics that relay frames between the n port switches. And a plurality of sublinks respectively connecting the n port switches and the m fabric switches. Each of the n port switches sets a link aggregation group with respect to m sublinks respectively connecting between itself and m fabric switches. Here, each of the n port switches transmits a broadcast frame for address resolution on (m−1) sublinks excluding the first sublink which is one of the m sublinks. When received, if the transmission source address matches the address of any one of the m fabric switches, the relay of the frame is permitted, and if it does not match, the frame is discarded.

  Of the inventions disclosed in the present application, the effects obtained by the representative embodiments will be briefly described. In the switch device and the network relay system including the switch device, improvement in maintainability can be realized.

In the network relay system by one embodiment of this invention, it is a block diagram which shows the structural example and the operation example used as a premise. It is the schematic which shows an example of the address table synchronization function with which the network relay system of FIG. 1 is provided. It is the schematic which shows an example of the in-band management function with which the network relay system of FIG. 1 is provided. It is the schematic which shows an example of the problem which arises when performing in-band management in the state which mounts the address table synchronization function of FIG. It is the schematic which shows an example of the solution system with respect to the problem of FIG. FIG. 2 is a flowchart showing an example of schematic processing contents of the port switch in the network relay system of FIG. 1. FIG. 2 is a block diagram illustrating a schematic configuration example of a main part of the port switch in the network relay system of FIG. 1. (A) is the schematic which shows the structural example of the address table in FIG. 7, (b) is the schematic which shows the structural example of the LAG table in FIG. FIG. 8 is a schematic diagram illustrating a structure example of an exception table in FIG. 7. FIG. 8 is an explanatory diagram illustrating an operation example of an exception table setting unit in FIG. 7. It is the schematic which shows an example of the setting content different from FIG. 9 of the exception table in FIG.

  In the following embodiment, when it is necessary for the sake of convenience, the description will be divided into a plurality of sections or embodiments. However, unless otherwise specified, they are not irrelevant, and one is the other. Some or all of the modifications, details, supplementary explanations, and the like are related. Further, in the following embodiments, when referring to the number of elements (including the number, numerical value, quantity, range, etc.), especially when clearly indicated and when clearly limited to a specific number in principle, etc. Except, it is not limited to the specific number, and may be more or less than the specific number.

  Further, in the following embodiments, the constituent elements (including element steps and the like) are not necessarily indispensable unless otherwise specified and apparently essential in principle. Needless to say. Similarly, in the following embodiments, when referring to the shapes, positional relationships, etc. of the components, etc., the shapes are substantially the same unless otherwise specified, or otherwise apparent in principle. And the like are included. The same applies to the above numerical values and ranges.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that components having the same function are denoted by the same reference symbols throughout the drawings for describing the embodiment, and the repetitive description thereof will be omitted.

<< Configuration and prerequisite operation of network relay system >>
FIG. 1 is a block diagram showing a configuration example and a premise operation example in a network relay system according to an embodiment of the present invention. The network relay system shown in FIG. 1 includes a box type fabric system 10. The box-type fabric system 10 includes n (n is an integer of 2 or more) port switches PS [1] to PS [n] and m (m is an integer of 2 or more) fabric switches FS [1] to FS. [M] and a plurality of sublinks (for example, SL [1, 1], etc.) that respectively connect n port switches and m fabric switches. The m fabric switches FS [1] to FS [m] are responsible for relaying frames between the n port switches PS [1] to PS [n].

  Each of the n port switches PS [1] to PS [n] and the m fabric switches FS [1] to FS [m] is configured by a box-type switch device. Each of the n port switches PS [1] to PS [n] includes m fabric switch ports Pf [1] to Pf [m] and p (p is an integer of 1 or more) user ports. Pu [1] to Pu [p]. Each of the m fabric switches FS [1] to FS [m] includes n port switch ports Pp [1] to Pp [n].

  Each of the n port switches PS [1] to PS [n] performs LAG with respect to m sublinks respectively connecting between itself and m fabric switches FS [1] to FS [m]. Set. For example, the port switch PS [1] includes m sublinks SL [1,1], SL [1,2],..., SL between m fabric switches FS [1] to FS [m]. LAG is set for [1, m]. Here, the LAG between the m sublinks is referred to as a main link ML [1]. Similarly, the port switch PS [2] includes m sublinks SL [2,1], SL [2,2],... Between the m fabric switches FS [1] to FS [m]. LAG (that is, main link ML [2]) is set for SL [2, m]. Similarly, the port switch PS [n] transmits LAG (that is, main link ML [n] to m sublinks SL [n, 1], SL [n, 2],..., SL [n, m]. ]).

  Hereinafter, each of the port switches PS [1] to PS [n] is referred to as a port switch PS, and each of the fabric switches FS [1] to FS [m] is referred to as a fabric switch FS. Each of the fabric switch ports Pf [1] to Pf [m] is referred to as a fabric switch port Pf, and each of the user ports Pu [1] to Pu [p] is represented as a user port. Each of the port switch ports Pp [1] to Pp [n] is referred to as a port switch port Pp. Further, each of the sublinks SL [1,1] to SL [n, m] is referred to as a sublink SL, and each of the main links ML [1] to ML [n] is represented as a main link ML. Called.

  Further, in the example of FIG. 1, each sublink SL is configured by two links 11. The link 11 means an aggregate including a communication line and ports at both ends thereof (that is, the fabric switch port Pf and the port switch port Pp). Each port switch (for example, PS [1]) sets a LAG for the two links 11 constituting each sublink (for example, SL [1, 1]). That is, the LAG is a LAG within a sublink, unlike the LAG between sublinks (that is, the main link ML) described above.

  Similarly, each fabric switch (for example, FS [1]) also sets LAG for the two links 11 constituting each sublink (for example, SL [1, 1]). Each sublink SL does not necessarily need to be composed of two links 11, and may be composed of one link 11 or three or more links 11 for each sublink SL. A LAG in the sublink is set in the sublink SL configured by two or more links 11.

  Here, for example, a frame is directed from the terminal TM1 connected to the user port Pu [1] of the port switch PS [1] to the terminal TM4 connected to the user port Pu [p] of the port switch PS [2]. Assume that FL1 is transferred. In this case, when the port switch PS [1] receives the frame FL1, the port switch PS [1] determines the relay destination fabric switch port Pf (in other words, the sublink SL) based on a predetermined distribution rule. In this example, a fabric switch port Pf [1] is defined. As a result, the frame FL1 is transferred to the port switch PS [2] and the terminal TM4 through a route passing through the fabric switch FS [1].

  Also, the frame FL2 is transferred from the terminal TM2 connected to the user port Pu [p] of the port switch PS [1] to the terminal TM3 connected to the user port Pu [1] of the port switch PS [2]. Assuming that In this case, when the port switch PS [1] receives the frame FL2, the port switch PS [1] determines the relay destination fabric switch port Pf (in other words, the sublink SL) based on a predetermined distribution rule. In this example, the fabric switch port Pf [2] is defined. As a result, the frame FL2 is transferred to the port switch PS [2] and the terminal TM3 through a route passing through the fabric switch FS [2]. If there is a failure in the fabric switch port Pf [2], other fabric switch ports Pf are determined.

  As described above, when the box-type fabric system 10 is used, load distribution and redundancy within the LAG can be realized along with the LAG. For example, when it is desired to expand the communication band, the fabric switch FS may be added, and the communication band can be easily expanded at low cost. Further, by adding the port switch PS, the number of ports (that is, the user port Pu) can be easily expanded at low cost. As a result, when this system is used, it is possible to construct a flexible system according to the user's request at a lower cost than when using a system comprising a chassis type switch device.

<Outline of address table synchronization function>
FIG. 2 is a schematic diagram illustrating an example of an address table synchronization function provided in the network relay system of FIG. FIG. 2 shows a partial configuration example when the number of fabric switches FS “m” is three in the box-type fabric system 10 of FIG. 1. In FIG. 2, for example, the terminal TM1 connected to the user port Pu [1] of the port switch PS [1] obtains an unknown MAC (Media Access Control) address from a known IP (Internet Protocol) address. An ARP request frame AF1 may be transmitted to perform address resolution.

  In the ARP request frame AF1, “FF... FFh” indicating a broadcast address is stored in the destination MAC address (DMAC) 25, and the MAC address (here, MA1) of the terminal TM1 is stored in the source MAC address (SMAC) 26. Is stored. The frame type 27 stores “0806h” indicating ARP, and the data 28 stores an IP address for address resolution.

  Here, if the address table synchronization function as shown in Patent Document 1 is not installed, the port switch PS [1] adds the ARP request frame AF1 to another user port Pu, and adds a LAG (ie, main). Along with the link ML [1]), flooding is performed to any one of the fabric switch ports (for example, Pf [1]). On the other hand, when the address table synchronization function is installed, the port switch PS [1] relays the ARP request frame AF1 to all the fabric switches FS [1] to FS [3]. In the present specification, this function is referred to as a LAG ARP exception function. Thereby, each fabric switch FS [1] -FS [3] can learn MAC address (MA1) of terminal TM1 simultaneously.

  However, when such an LAG ARP exception function is provided, each port switch PS may simultaneously receive ARP request frames having the same contents from all the fabric switches FS [1] to FS [3]. In the example of FIG. 2, the port switch PS [2] (also PS [3] to PS [n]) receives the ARP request frame AF1 from all the fabric switches FS [1] to FS [3]. Then, the port switch PS [2] needs to process the same three ARP request frames AF1 from the fabric switches FS [1] to FS [3].

  Therefore, in order to prevent such a situation, each port switch PS has a function of discarding an ARP request frame received from other than a predetermined fabric switch (for example, FS [1]) in addition to an exception function at the time of ARP of LAG. Have. In this specification, this function is called an ARP discard function. Specifically, in the port switch PS [2] (PS [1], PS [3] to PS [n] are also the same), for example, the ARP received by the fabric switch ports Pf [2] and Pf [3]. Condition setting 15 is performed to discard the request frame AF1.

  Accordingly, even when the port switch PS [2] (PS [1], PS [3] to PS [n] is the same) receives the same ARP request frame from the fabric switches FS [1] to FS [3]. , Processing can be performed only on the ARP request frame received from the fabric switch FS [1]. In the example of FIG. 2, the port switch PS [2] (also PS [3] to PS [n]) floods the ARP request frame AF1 received at the fabric switch port Pf [1] to the user port Pu. .

  By performing the processing as described above, the ARP request frame AF1 can be transferred to each terminal connected to each port switch PS. The address table synchronization function is configured by a combination of the LAG ARP exception function and the ARP discard function described above.

<Outline of in-band management>
FIG. 3 is a schematic diagram illustrating an example of an in-band management function provided in the network relay system of FIG. FIG. 3 shows a partial configuration example when the number of fabric switches FS “m” is three in the box-type fabric system 10 of FIG. 1. In the in-band management, for example, as shown in FIG. 3, the management terminal MTM1 is connected to the user port Pu [1] of the port switch PS [1], and the maintenance of the fabric switch FS [2] is performed from the management terminal MTM1.・ Management is performed. Examples of maintenance / management include information collection, various settings, firmware update, and the like.

  At this time, the port switch PS [1] sets the LAG (that is, the main link ML [1]) to the sublink SL between the fabric switches FS [1] to FS [3]. Communication between the terminal MTM1 and the fabric switch FS [2] may not be performed normally. That is, since the frame FL3 transmitted from the management terminal MTM1 is distributed according to a predetermined rule by the LAG (main link ML [1]) of the port switch PS [1], it does not necessarily reach the fabric switch FS [2]. Not always.

  Therefore, in order to prevent such a situation, each port switch PS (here, PS [1]) includes a condition table 20, for example. Each port switch PS [1] has a mechanism that gives priority to the condition table 20 over LAG. In the condition table 20, the destination MAC address is the MAC address of the fabric switch FS [2] (in this case, MAf2), and the identifier (port ID) {Pf [2]} of the relay destination port in that case It is shown. Here, {Pf [2]} means a port ID assigned to the port Pf [2]. Similarly, in the present specification, for example, {AA} means an identifier (ID) attached to AA.

  Thereby, the frame FL3 transmitted from the management terminal MTM1 can reach the fabric switch FS [2] without being distributed by the port switch PS [1]. That is, the frame FL3 includes the MAC address (MAf2) of the fabric switch FS [2] as the destination MAC address (DMAC) 25. Accordingly, when the port switch PS [1] receives the frame FL3, the destination MAC address of the frame FL3 and the destination MAC address set in the condition table 20 match, so the frame FL3 is selected based on the condition table 20. It can be relayed to the fabric switch port Pf [2]. Conversely, a frame transmitted from the fabric switch FS [2] and destined for the MAC address (MA2 in this case) of the management terminal MTM1 is sent to the in-band management function except for the case described in FIG. Regardless, the management terminal MTM1 can be reached.

《In-band management with the address table synchronization function (problem)》
FIG. 4 is a schematic diagram showing an example of problems that occur when performing in-band management in a state where the address table synchronization function of FIG. 2 is installed. For example, during in-band management, the fabric switch FS may transmit an ARP request frame with itself as a transmission source for address resolution. Although not particularly limited, for example, in the process in which management communication is performed between the management terminal MTM1 and the fabric switch FS [2] as shown in FIG. 3, the management terminal is switched from the fabric switch FS [2]. Address resolution may be required when the MTM1 MAC address has disappeared.

  As shown in FIG. 4, when the fabric switch FS [2] transmits the ARP request frame AF3 by broadcast, the port switch PS [1] (also PS [2] to PS [n]) transmits the ARP request frame AF3. Is received by the fabric switch port Pf [2]. However, when the port switch PS [1] (also PS [2] to PS [n]) has an address table synchronization function, the fabric switch port Pf [2] is accompanied with the condition setting 15 described in FIG. The ARP request frame AF3 received in step 1 is discarded. As a result, the ARP request frame AF3 cannot reach each terminal including the management terminal MTM1, and sufficient in-band management may not be realized.

<< In-band management function (solution) with address table synchronization function >>
FIG. 5 is a schematic diagram showing an example of a solution method for the problem of FIG. In order to solve the problem shown in FIG. 4, the port switch PS [1] (PS [2] to PS [n] is similar) includes, for example, a reception permission table 30 as shown in FIG. In the reception permission table 30, the target port ID is {Pf [2]}, {Pf [3]}, and the MAC addresses of the source MAC addresses of the fabric switches FS [2], FS [3] It is shown that the address (here, MAf2 and MAf3) and the frame type is an ARP request. The port switch PS [1] permits relaying of frames that satisfy the conditions of the reception permission table 30 from the received frames.

  As a result, the ARP request frame AF3 transmitted from the fabric switch FS [2] is not discarded by the condition setting 15 of the port switch PS [1] (the same applies to PS [2] to PS [n]) and is used for management. Each terminal can be reached including the terminal MTM1. More specifically, first, in the ARP request frame AF3, the broadcast address as the destination MAC address (DMAC) 25, the MAC address (MAf2) of the fabric switch FS [2] as the source MAC address (SMAC) 26, As the frame type 27, an ARP identifier is included.

  Therefore, when the port switch PS [1] receives the ARP request frame AF3 at the fabric switch port Pf [2], the port switch PS [1] sets various information in the received port and the ARP request frame AF3 and the reception permission table 30. Therefore, the relay of the ARP request frame AF3 is permitted. In this case, the condition in the reception permission table 30 is prioritized over the condition setting 15. As a result, the port switch PS [1] (same for PS [2] to PS [n]) floods the ARP request frame AF3 to the user port Pu. As a result, the ARP request frame AF3 can reach each terminal connected to each port switch PS including the management terminal MTM1.

  As described above, by using the in-band management function of FIG. 5, it is possible to realize sufficient in-band management even when the address table synchronization function is installed, in addition to the in-band management function of FIG. It becomes possible. As a result, system maintainability can be improved. In particular, in a box-type fabric system, each switch is physically distributed as appropriate. For this reason, for example, in the above-described method of providing a dedicated port for maintenance / management (management port), there is a risk of increasing the communication line, or communication for maintenance / management when a fabric switch is added. It may be necessary to construct a route separately. By realizing in-band management by the method of this embodiment, it is possible to improve maintainability including the solution of such problems.

<Overview of port switch operation>
FIG. 6 is a flowchart showing an example of schematic processing contents of the port switch in the network relay system of FIG. FIG. 6 shows a part of the frame reception processing performed when a frame is received at the port, and reflects the address table synchronization function shown in FIG. 2 and the in-band management function shown in FIG. It has become a flow. The port switch PS first analyzes the header of the received frame and determines whether or not the frame is an ARP request frame that becomes a broadcast frame (step S101). If it is an ARP request frame, the process of step S102 is performed. If it is not an ARP request frame, the process of step S106 is performed.

  In step S106, the port switch PS completes the frame reception process after performing a predetermined process as a representative of a normal relay process based on the address table. On the other hand, in step S102, the port switch PS determines whether or not the ARP request frame is received by the sublink SL (in other words, the fabric switch port Pf). In this example, when not received by the sublink SL, it is received by the user port Pu. After the port switch PS floods the ARP request frame to the other user ports Pu and all the sublinks FL, The frame reception process ends (step S107). This processing is performed by the exception function at the ARP of LAG described in FIG.

  On the other hand, in step S102, when received by the sublink SL, the port switch (for example, PS [1]) receives the sublink (SL [1,1]) excluding the predetermined sublink (for example, SL [1,1]). 2] to SL [1, m]), it is determined (step S103). When it is received by the predetermined sublink (SL [1, 1]), the port switch (PS [1]) permits the relay of the ARP request frame (step S105). On the other hand, when received on sublinks (SL [1,2] to SL [1, m]) excluding a predetermined sublink, the port switch (PS [1]) has a source MAC address of the fabric switch (FS [2] to FS [m]) is determined (step S104).

  In step S104, in the case of the MAC address of the fabric switch FS, the port switch PS permits the relay of the ARP request frame (step S105). Thereafter, the port switch PS performs a predetermined process, and then ends the frame reception process (step S106). Specifically, the port switch PS floods the ARP request frame to each user port Pu. This processing is performed by the in-band management function described in FIG. On the other hand, if it is not the MAC address of the fabric switch FS in step S104, the port switch PS terminates the frame reception process after discarding the ARP request frame (step S108). This processing is performed by the ARP discard function described in FIG.

  As described above, each port switch (for example, PS [1]) is a first sublink that is one of m sublinks (SL [1,1] to SL [1, m]). When an ARP request frame is received on (m−1) sublinks (SL [1,2] to SL [1, m]) excluding (SL [1,1]) (step S103), The following cases are classified. That is, each port switch PS permits the relay of the ARP request frame when the source address of the ARP request frame matches the address of any one of the m fabric switches FS. , The ARP request frame is discarded (steps S104, S105, S108). When each port switch (for example, PS [1]) receives the ARP request frame on the first sublink (SL [1, 1]) (step S103), it permits relaying of the ARP request frame (step S103). Step S105).

<Details of port switch>
FIG. 7 is a block diagram illustrating a schematic configuration example of a main part of the port switch in the network relay system of FIG. 7 includes, for example, a frame processing unit 35, a table unit 36, an exception table setting unit 37, a plurality of ports (user ports Pu [1] to Pu [p], and fabric switch ports). Pf [1] to Pf [m]). A terminal or the like is appropriately connected to the user ports Pu [1] to Pu [p] via a communication line.

  The fabric switch ports Pf [1] to Pf [m] are connected to m fabric switches FS [1] to FS [m] via sublinks SL [1] to SL [m], respectively. In this example, each of the sublinks SL [1] to SL [m] is composed of two links 11, and accordingly, each of the fabric switch ports Pf [1] to Pf [m] Consists of two fabric switch ports. For example, the fabric switch port Pf [1] is composed of the fabric switch ports Pf [1,1] and Pf [1,2], and the fabric switch port Pf [m] is the fabric switch port Pf [m]. , 1], Pf [m, 2].

  As described in FIG. 1, LAGs between sublinks (that is, main link ML [1]) are set in sublinks SL [1] to SL [m]. Moreover, LAG in a sublink is set with respect to each of sublink SL [1] -SL [m].

  The table unit 36 includes an address table 45, a LAG table 46, and an exception table 47. The address table 45 is a table showing the relationship between each port and the MAC address existing ahead of each port. The LAG table 46 is a table showing the correspondence relationship between the main link ML [1], the sublinks SL [1] to SL [m], and the port (here, the fabric switch port Pf). The LAG table 46 can be generated in a fixed manner by an administrator or the like in advance, or can be automatically generated using a method such as that disclosed in Patent Document 2.

  The exception table 47, which will be described in detail later, implements the function of the condition table 20 shown in FIG. 3, the function of the reception permission table 30 shown in FIG. 5, or the address table synchronization function described in FIG. It is a table to do. Although the details will be described later, the exception table setting unit 37 automatically sets the exception table 47.

  The frame processing unit 35 includes a LAG distribution control unit 40 and an exception table execution unit 41. The frame processing unit 35 mainly performs a process of searching for a destination port using the address table 45 when a frame is received at each port and relaying the received frame to the port. At this time, for example, when the destination port of the frame received at the user port Pu is the port in which the main link ML [1] is set (that is, the fabric switch port Pf), the frame processing unit 35 Using the distribution control unit 40, the relay destination of the frame is appropriately distributed in the main link ML [1]. Further, the exception table execution unit 41 performs processing on the frame based on the exception table 47.

  FIG. 8A is a schematic diagram showing an example of the structure of the address table in FIG. 7, and FIG. 8B is a schematic diagram showing an example of the structure of the LAG table in FIG. The LAG table 46 shown in FIG. 8B indicates that the main link identifier (main link ID) {ML [1]} is composed of sublink IDs {SL [1] to SL [m]}. In other words, it is indicated that the main link ML [1] is composed of sub links SL [1] to SL [m]. Further, in the LAG table 46, each port ID corresponding to each sublink ID {SL [1] to SL [m]} is shown. For example, the sublink ID {SL [1]} is composed of the port IDs {Pf [1,1], Pf [1,2]}. In other words, the sublink SL [1] is composed of fabric switch ports Pf [1,1] and Pf [1,2].

  In this example, the LAG table 46 also indicates the state of each link 11 corresponding to each of the fabric switch ports Pf [1, 1] to Pf [m, 2] (that is, whether there is a failure). For example, if the frame processing unit 35 in FIG. 7 detects a failure in the fabric switch ports Pf [1, 1] to Pf [m, 2] by periodically transmitting / receiving control frames, the frame processing unit 35 displays the information as LAG. Record in table 46. When a failure is detected, the LAG distribution control unit 40 recognizes the link 11 having the failure based on the LAG table 46, and performs the distribution process except for the link 11 having the failure.

  The address table 45 shown in FIG. 8A shows the relationship between the port ID / main link ID and the MAC address existing ahead of the port corresponding to the port ID / main link ID. For example, when the destination MAC address of the frame received at any port is “MAXx”, the frame processing unit 35 in FIG. 7 relays the frame to the user port Pu [1] based on FIG. To do. Further, when the destination MAC address of the frame received at any of the user ports Pu [1] to Pu [p] is “MAzz”, the frame processing unit 35 sets the frame as the main based on FIG. Relay to link ML [1]. At this time, the LAG dispersion control unit 40 operates.

<Details of exception table>
FIG. 9 is a schematic diagram showing an example of the structure of the exception table in FIG. In the exception table 47a shown in FIG. 9, processing conditions including conditions (in this example, conditions at the time of reception) and permission or cancellation of relaying a frame that satisfies the conditions are set. In the condition, a port to be processed and information to be compared with information in a frame received at the port are set. The information to be compared is, for example, various types of information included in the Ethernet header or IP header, with the source / destination MAC address, frame type, source / destination IP address, etc. as representatives. With regard to the processing contents, in addition to setting whether to permit or discard the relay of the frame, it is possible to specify / change the relay destination of the frame.

  For example, when the processing content described in FIG. 5 is realized, the ports to be processed are (m−1) ports excluding the sublink SL [1], as indicated by “No. 1” in FIG. The fabric switch ports Pf [2] to Pf [m] corresponding to the sublinks SL [2] to SL [m]. The information to be compared is a source MAC address, a frame type, and a destination MAC address. The processing content is permission to relay the frame.

  In the transmission source MAC address, addresses of (m−1) fabric switches FS [2] to FS [m] corresponding to (m−1) sublinks are described. In the frame type, an identifier (specifically “0806h”) indicating an ARP command is described. In the destination MAC address, a broadcast address (specifically “FF... FFh”) is described.

  When the exception table execution unit 41 performs processing based on such an exception table 47a, an in-band management function as shown in FIG. 5 can be realized. That is, each port switch PS uses the fabric switch ports Pf [2] to Pf as ARP request frames having the MAC addresses of the (m−1) fabric switches FS [2] to FS [m] as the source MAC addresses. When received at [m], the ARP request frame can be relayed to the user port Pu.

  Actually, the in-band management function and the address table synchronization function (ARP discard function) described above must be used in combination. In this case, a mechanism may be used in which the in-band management function is prioritized over the ARP discard function. The method for realizing the ARP discard function is not particularly limited. For example, the ARP discard function can be realized by “No. 2” of the exception table 47a in FIG.

  In the exception table 47a of FIG. 9, the smaller “No.” is set to a higher priority. In “No. 2” of the exception table 47a, the ports to be processed are the same fabric switch ports Pf [2] to Pf [m] as “No. 1”. The information to be compared is a frame type (ARP) indicating a ARP request frame and a destination MAC address (broadcast address). The processing content is the discard of the frame.

  Thus, for example, when an ARP request frame having a source MAC address that is not the MAC address of the fabric switch FS is received by the fabric switch port Pf [2], “No. 2” is applied to the ARP request frame. “No. 1” is not applied. As a result, the ARP request frame is discarded. On the other hand, when the ARP request frame having the MAC address of the fabric switch FS as the transmission source MAC address is received by the fabric switch port Pf [2], the ARP request frame includes “No. 2” and “No. 1”. Both apply. However, since “No. 1” has a higher priority, the ARP request frame is permitted to be relayed and relayed to the user port Pu.

<Operation of exception table setting section>
FIG. 10 is an explanatory diagram showing an operation example of the exception table setting unit in FIG. FIG. 10 shows a partial configuration example in the case where the number of fabric switches FS “m” is three in the box-type fabric system 10 of FIG. 1. As illustrated in FIG. 10, each of the fabric switches FS [1] to FS [3] periodically detects from the port switch ports Pp [1] to Pp [n] in order to detect a failure of each link 11, for example. A control frame is transmitted.

  In this example, the fabric switch FS [1] transmits the control frame CF [1] from the port switch ports Pp [1] to Pp [n], and similarly, the fabric switches FS [2] and FS [3]. Respectively generate control frames CF [2] and CF [3]. Each control frame CF [1] to CF [3] includes a source MAC address (SMAC) 26. For example, the control frame CF [2] includes the MAC address (in this case, MAf2) of the fabric switch FS [2].

  Therefore, the exception table setting unit 37 of the port switch PS [1] (same for PS [2] to PS [n]) receives the control frames CF [1] to CF [3] and includes the fabric included in them. The exception table 47a as shown in FIG. 9 is automatically set using the MAC addresses of the switches FS [1] to FS [3]. That is, as can be seen from FIG. 9, if the MAC address of each fabric switch port Pf and each fabric switch FS existing ahead is known, the remaining information is fixed information, and thus such automatic setting is possible. Become. The exception table setting unit 37 is realized by program processing using a CPU (Central Processing Unit), for example.

<< Details of exception table (application example) >>
FIG. 11 is a schematic diagram showing an example of setting contents different from FIG. 9 of the exception table in FIG. For example, the ARP exception function of LAG described in FIG. 2 and the in-band management function described in FIG. 3 can be realized by the exception table 47b as shown in FIG. As in the case of FIG. 9, the exception table 47b shown in FIG. 11 sets conditions (in this example, conditions at the time of reception) and processing contents for frames that satisfy the conditions. However, here, it is possible to specify and change the relay destination of the frame as the processing content.

  The settings of “No. 1” to “No. m” in FIG. 11 correspond to the in-band management function in FIG. For example, in “No. 1”, user ports Pu [1] to Pu [p] are set as processing target ports, and a destination MAC address is set as information to be compared. The destination MAC address describes the MAC address of the fabric switch FS [1]. The processing content is set to change the relay destination to a predetermined sublink to sublink SL [1].

  Similarly, in “No. m”, user ports Pu [1] to Pu [p] are set as processing target ports, and a destination MAC address is set as information to be compared. The destination MAC address describes the MAC address of the fabric switch FS [m]. The processing content is set to change the relay destination to a predetermined sublink to the sublink SL [m]. As a result, each port switch PS correctly relays the frame toward the fabric switch FS when the user MAC port Pu receives a frame in which any MAC address of the fabric switch FS is described as the destination MAC address. It becomes possible.

  Further, the setting of “No. k” in FIG. 11 corresponds to the exception function at the time of ARP of LAG in FIG. In “No. k”, the user ports Pu [1] to Pu [p] are set as the processing target ports, and the frame type (ARP) indicating the ARP request frame and the destination MAC are included in the information to be compared. An address (broadcast address) is set. The processing content is set to change the relay destination to a predetermined sublink to all sublinks SL [1] to SL [m]. Thus, when each port switch PS receives the ARP request frame at the user port Pu, the port switch PS can relay the frame toward all the fabric switches FS [1] to FS [m].

  Although not particularly limited, the exception table execution unit 41 of FIG. 7 is installed at the subsequent stage of the LAG distribution control unit 40, and a predetermined sublink determined as a destination by the LAG distribution control unit 40 is based on the exception table 47b. Perform the change process. The exception table 47 and the exception table execution unit 41 can be realized by applying a function called a packet filter, for example. The packet filter is installed as a security function (so-called firewall function), refers to predetermined information in the frame (for example, destination address, source address, etc.), and the referenced information matches information in a predetermined table. This is a function that determines whether or not the frame can be passed depending on whether or not it is performed.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiments. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the invention. For example, the above-described embodiment has been described in detail for easy understanding of the present invention, and is not necessarily limited to one having all the configurations described. Further, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. . Further, it is possible to add, delete, and replace other configurations for a part of the configuration of each embodiment.

  For example, although an ARP request frame has been described here as an example, similarly, any ARP request frame may be applied to any address resolution broadcast frame. Specifically, for example, contrary to the ARP request frame, there is a RARP request frame that resolves an IP address from a MAC address, or a “Neighbor Solitization (NS) frame” that corresponds to the IPv6 version of the ARP request frame. It is done.

DESCRIPTION OF SYMBOLS 10 Box type fabric system 11 Link 15 Condition setting 20 Condition table 25 Destination MAC address 26 Source MAC address 27 Frame type 28 Data 30 Reception permission table 35 Frame processing part 36 Table unit 37 Exception table setting part 40 LAG distribution control part 41 Exception table execution unit 45 Address table 46 LAG table 47, 47a, 47b Exception table AF1, AF3 ARP request frame CF [1] to CF [3] Control frame FL1, FL2, FL3 frame FS Fabric switch ML Main link MTM1 Management terminal PS port switch Pf Fabric switch port Pp Port switch port Pu User port SL Sublink TM1 to TM4 terminal

Claims (10)

n port switches (n is an integer of 2 or more);
M (m is an integer of 2 or more) fabric switches responsible for relaying frames between the n port switches;
A plurality of sublinks respectively connecting the n port switches and the m fabric switches;
With
Each of the n port switches sets a link aggregation group for m sublinks that connect between itself and the m fabric switches, respectively.
Each of the n port switches is
When a broadcast frame for address resolution is received on (m−1) sublinks excluding any one of the m sublinks, the broadcast frame transmission source When the address matches the address of any one of the m fabric switches, the relay of the broadcast frame is permitted, and when the address does not match, the broadcast frame is discarded.
Network relay system. The network relay system according to claim 1,
Each of the n port switches further permits relaying of the broadcast frame when the broadcast frame is received on the first sublink. The network relay system according to claim 2,
Each of the n port switches is
An exception table in which processing conditions including permission and discard of relay for frames satisfying the condition and the condition are set;
An exception table execution unit that performs processing on a frame based on the exception table;
With
In the condition, a port to be processed and information to be compared with information in a frame received at the port are set.
The port to be processed is a port corresponding to the (m−1) sublinks,
The information to be compared includes a frame type in which an identifier representing an ARP command is described, and addresses of (m−1) fabric switches corresponding to the (m−1) sublinks. Source address,
Network relay system. The network relay system according to claim 3, wherein
Each of the n port switches receives a control frame periodically transmitted from each of the m fabric switches, and uses each of the addresses of the m fabric switches included in the control frame. A network relay system that automatically sets the exception table. In the network relay system according to any one of claims 1 to 4,
The network relay system, wherein the address resolution broadcast frame is an ARP request frame. A switch device that is connected to each of m (m is an integer of 2 or more) fabric switches via m sublinks and operates by setting a link aggregation group for the m sublinks. ,
When a broadcast frame for address resolution is received on (m−1) sublinks excluding any one of the m sublinks, the broadcast frame transmission source When the address matches the address of any one of the m fabric switches, the relay of the broadcast frame is permitted, and when the address does not match, the broadcast frame is discarded.
Switch device. The switch device according to claim 6, wherein
Furthermore, when the broadcast frame is received on the first sub-link, the switching device permits relaying the broadcast frame. The switch device according to claim 7, wherein
An exception table in which processing conditions including permission and discard of relay for frames satisfying the condition and the condition are set;
An exception table execution unit that performs processing on a frame based on the exception table;
Have
In the condition, a port to be processed and information to be compared with information in a frame received at the port are set.
The port to be processed is a port corresponding to the (m−1) sublinks,
The information to be compared includes a frame type in which an identifier representing an ARP command is described, and addresses of (m−1) fabric switches corresponding to the (m−1) sublinks. Source address,
Switch device. The switch device according to claim 8, wherein
In addition, the control frame periodically transmitted from each of the m fabric switches is received, and the exception table is automatically set using the addresses of the m fabric switches included in the control frame. A switching device having an exception table setting unit. The switch device according to any one of claims 6 to 9,
The switch device, wherein the address resolution broadcast frame is an ARP request frame.

Images