JP2612053B2 - Degradation method in case of failure - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a connection to a local network (LAN) and a LAN serving as an intermediary function for communication between end systems ES accommodated in a wide area network (WAN). SCSI interface for information transfer between the access device AU and the gateway device GW connected to the WAN, degenerates to a normal system to avoid communication interruption even when a failure occurs in the connection unit CU on the gateway device GW side The present invention relates to a method (a method of enabling a normal operation in a form in which a failed unit is separated and the remaining units are reduced).

[Conventional technology]

From the viewpoint of the end system ES in the LAN, if a plurality of connection units CU of the gateway device GW are accommodated for improving the performance and reliability, the LA is provided for each connection unit CU.
It is configured equivalently to one having a plurality of gateway devices GW to which N addresses are assigned.

If the degeneration function at the time of failure is not provided as in the conventional configuration in this configuration, the LAN end system ES uses the destination address of the predetermined gateway device GW as the destination address, and when communication cannot be performed with this, the destination address of another gateway device GW is used. And re-communicated. That is, in the recall processing of the end system ES in the LAN, the destination address needs to be changed.

[Problems to be solved by the invention]

In the present invention, the connection system CU of the gateway device GW is duplicated, and when the connection unit CU fails, the relay system of the access device AU and the gateway device GW is degraded to one of the normal sides, so that the end system ES in the LAN is degraded. Is a problem to be solved so that special processing (recall with a changed address) is not required to secure communication with the end system ES of the WAN. The purpose is to provide a degeneration method.

[Means for solving the problem]

In order to solve the above-mentioned problem, in the present invention, a SCSI interface as a U.S. national standard already known per se is used as an interface between a LAN access device and a gateway device GW.

Here, ANSI X3.131-1986 “Small Computer System
Interface (SCSI) ”means ANSI (American Standards Association)
The X3 committee, which had been studying X3T9.2, was approved by ANSI through formal procedures in June 1986, and was enacted as an American national standard.

SCSI is a peripheral device interface for small computer systems. Compared to conventional devices, SCSI not only incorporates the functions of the controller, but also allows up to eight devices to be connected and allows communication between controllers without the intervention of a host. The major feature is that it is possible. As a result, multitask processing can be easily realized, and a great merit can be obtained also in software development.

A first purpose of this interface is to provide a device independent host computer for a device belonging to one model. Thus, different disk drives, tape drives, printers, and even communications devices can be connected without the need for common system hardware and software modifications. The addition of unusual features and functions is designed to be done with vendor-specific fields and codes.

This interface uses logical rather than physical addressing for all data blocks. The direct access device can be queried to know how many blocks each logical unit has. The logical unit corresponds to all or a part of the peripheral device.

The interface protocol includes a plurality of initiators (SCSI devices capable of starting an operation) and a plurality of targets (SCSI capable of responding to a request for performing an operation).
Device).

In short, SCSI is a general-purpose interface, in which a plurality of devices are connected on an interface line, and the times of the plurality of devices are largely identified by IDs (identifiers). This is an interface that can be further subdivided by using a LUN (logical unit number) inside the device correspondence of the ID. The present invention relates to such a general-purpose interface. Can be said to have been used skillfully.

[Action]

In a gateway device that relays between an end system in a local network and an end system in a wide area network, a connection between an access device and a gateway device in the local network is a SCSI interface, and a plurality of connection units on the gateway device side on the SCSI interface are provided. Configuration, and assigns a device identifier and a logical unit number to the access device and each connection unit on the SCSI interface. The device identifier uses a different value for the access device and each connection unit, and a different logical value for each connection unit. A unit number is assigned, and the information transfer between the access device and the gateway device is distinguished by the device identifier and the logical unit number. Each connection unit includes in advance the entire routing processing via the gateway device. On the gateway device side Re Is the connection unit when a failure in the access device is degenerated to disassociate and normal connection unit connecting unit of the fault by changing the device identifier of the connection unit of the failure to successfully device identifier.

〔Example〕

FIG. 1 shows a configuration example of an application system of the present invention as an embodiment of the present invention. In the figure, communication information in LAN is
Source SA and destination DA of LAN address for each communication frame FR
Is added at the beginning. The end system ES, the network management device NS, and the access device AU of each connection in the LAN pick up the destination LAN address DA of the received communication frame FR of the above-described format when it is determined that the destination LAN address DA matches that registered in the self. . Conversely, transmission to the LAN is performed by adding the transmission source SA and destination DA of the LAN address to the communication frame FR of each transmission.

The access device AU is connected to the gateway device GW via a LAN.
Responsible for transmission of internal and external communication information and electrical and physical processing for LAN. The gateway device GW has a WA in its processing unit PU.
N (dedicated line, switched line network, etc.), and is responsible for protocol processing and relay routing of communication information for communication with the connected end system ES.

The connection unit CU in the gateway device GW performs protocol and routing control for LAN. The network management device NS is responsible for the operation management of the LAN system, instructs the access device AU and the gateway device GW to start / end the operation, grasps the operation status thereof, and connects the gateway device GW of the present invention. When a failure occurs in the unit CU, the failure is grasped and a switching instruction is issued to the access device AU.

When the intra-LAN end system ES communicates with the WAN end system ES, the intra-LAN end system ES attaches a LAN address DA addressed to an access device AU corresponding to a predetermined gateway device GW to the communication frame FR and sends it out. This is taken by the access device AU and transferred to the gateway device GW via the SCSI interface. Conversely, gateway device G from WAN
When entering W, the gateway device GW adds the LAN address DA of the destination LAN end system ES to the communication information via the SCSI interface, and transmits the communication information to the access device AU. The access device AU prepares the communication frame FR and sends it to the end system ES in the LAN.

A SCSI interface is used between the access device AU and the gateway device GW. SCSI is defined as a method of transferring information between devices connected to the same bus. As shown in FIG. 2, at the beginning of the information transfer procedure, the communication direction is negotiated with the partner to which the information is transferred using the device identifier ID (SCSI
Then it is called arbitration, selection). Thereafter, in the command phase, the processing type in the connection unit CU is specified by the logical unit number LUN of the partner to which the data is transferred. After that, it is arranged to enter a data transfer phase and transfer communication information.

Device identifier ID is 0 to 7 for each connected device on the SCSI bus.
Up to eight types. Here, it is allocated to the access device AU and each connection unit CU. The logical unit number LUN is assigned from eight types of 0 to 7 in each connection device having the device identifier ID. Here, each connection unit CU uses a different value so that it can be distinguished from the original even in the case of degeneration. Also, depending on the type of information to be transferred (for example, the difference in communication protocol), the logical unit number LUN is used properly, and the AU LAN
Also use the internal address.

In the gateway device GW, the SCSI device identifier ID and the logical unit number LUN for each connection unit CU are, for example, as shown in FIG. 3A, the device identifier ID of the connection unit # 1 is 0, and the logical unit number LAN is 0 and 1. The device identifier ID of the connection unit # 2 is assigned to 1, and the logical unit number LUN is assigned to 2 and 3. In addition, this assignment also makes the address in the LAN of the access device AU, which is the destination address DA of the ES in the LAN, correspond to the address in FIG. 3A. As a result, from the viewpoint of the end system ES in the LAN, there are a plurality of LAN addresses for each communication information type for each connection unit CU, although there is only one gateway device GW itself. On the other hand, access device AU
Assigns, for example, 7 as a device identifier ID.

In the information transfer from the gateway device GW to the access device AU, a device identifier ID indicating the SCSI AU and a logical unit number
Information is transferred using the LUN. The information from the gateway device GW includes the LAN address DA of the destination ES in the LAN, but the source SA address in the LAN is obtained by extracting the LAN address from the logical unit number LUN corresponding to the access device AU. That is, the source address SA from the access device AU on the LAN to the end system ES is registered for reception.
Based on the correspondence table between the LAN address and the logical unit number LUN, the registered LAN address is extracted from the logical unit number LUN and used.

With this, the end system ES in LAN and the access device A
SA / DA on the communication frame FR with U corresponds. Therefore, it is necessary that the registered LAN address and the logical unit number LUN do not have the same combination of contents. That is, the logical unit number LUN corresponding to the LAN address does not use the same value in the same GW.

FIG. 4 shows a configuration example of the access device AU and the gateway device GW according to the present invention. The failure of each connection unit CU of the gateway device GW may be detected by any of the network management device NS, the access device AU, and the gateway device GW.

When implemented in the network management device NS, in addition to the relationship with the GW described later, the access device AU and the gateway device
The regular monitoring of the operating status of the GW allows the access device AU,
It is detected from the result of the response from the gateway device GW.

In the case of implementation in the access device AU, when receiving the communication frame FR from the LAN in the communication unit LC and passing it from the SCSI processing unit SC to the connection unit CU, it can be determined that it is not accepted by the gateway GW. . Further, the gateway device GW may notify the busy status by the SCSI control unit SCC. In this case, detection and subsequent switching of the connection unit CU can be performed only by the access device AU.

When the detection is performed by the connection unit CU of the gateway device GW, the processing status is monitored by the processing unit PU for each SCSI control unit SCC and the SCSI processing unit LU, and the result is monitored by the network management device.
Using one of the connection units CU to report to the NS (in the case of an abnormality, from the normal side), the network management apparatus NS manages the information, and the switching instruction from the network management apparatus NS to the access apparatus AU is issued. The processing performed by the gateway device GE can accurately determine the processing status.

Regardless of the detection method, there is no difference that the access device AU switches the connection unit CU corresponding to the LAN address of the connection unit CU that passes the communication information in order to switch with the connection unit CU of the gateway device GW.

This is as shown in the correspondence table TBN of the LAN address, the device identifier ID, and the logical unit number LUN of the access device AU, as exemplified in the case of the failure of the connection unit CU # 1 in FIG. 3B. The device identifier ID of the unit CU, that is, the device identifier ID corresponding to the LAN address of the gateway device GW may be changed from a faulty device to a normal connection unit CU.

Here, the logical unit number LUN to be changed is a number that makes the initial settings different between the connection units CU as described above, and is transmitted from the gateway apparatus GW to the end system E in the LAN.
The centralized management of the transmission source address SA in the access device AU at the time of transmission to S, and the connection unit CU of the gateway device GW can be distinguished from the initial one and the added one after the change.

The change of the device identifier ID registered in the processing table TB in the access device AU is performed when the detection of the access device AU itself is performed by a processing table of a correspondence list of the LAN address, the device identifier ID, and the logical unit number LUN in the information transfer process. Device identifier I of TB value
D can be rewritten. When the network management device NS performs detection management, the device identifier in the correspondence list of the LAN address, the device identifier ID, and the logical unit number LUN for the access device AU.
Download the ID change information and implement the change of the process table TB value.

In the gateway device GW, the switching of the connection unit CU is carried out at any time from the access device AU without the involvement of the gateway device GW. Relay processing and protocol processing for LAN.

〔The invention's effect〕

According to the present invention, when one of the connection units CU of the gateway device GW fails, the LA is dynamically changed during operation of the system.
Regardless of the end system ES in N, there is an advantage that degeneration can be realized by switching the device identifier ID of the SCSI interface between the connection unit CU of the access device AU and the gateway device GW to the connection unit CU on the normal side.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG.
FIG. 3A is a flowchart showing transfer control using a device identifier ID on the SCSI interface and a logical unit number LUN, and FIG. 3A shows the LAN address of the access device AU and the device identifier on the SCSI interface for each connection unit CU on the gateway device GW side. FIG. 3 (B) is an explanatory diagram showing a correspondence relationship between an ID and a logical unit number LUN, FIG. 3 (B) is an explanatory diagram showing a correspondence relationship when a failure occurs in a connection unit CU # 1, and FIG. 4 is a configuration of an access device AU and a gateway device GW. It is a block diagram showing an example. Explanation of symbols AU: access device, NS: network management device, ES
…… End system, LAN …… Local area network, GW …… Gateway device, CU …… Connection unit,
PU: Processing unit, WAN: Wide area network, LC: Communication unit, SC: S
CSI processing unit, TB ... processing table, SCC ... SCSI control unit,
LU: SCSI processing unit.

Continuation of the front page (56) References JP-A-63-131636 (JP, A) JP-A-62-280947 (JP, A) JP-A-63-90239 (JP, A) JP-A-63-138832 (JP, A) , A)

Claims (1)

(57) [Claims] In a gateway device GW that performs a relay routing between an end system ES in a local network LAN and an end system ES in a wide area network WAN, a connection between the LAN access device AU and the gateway device GW is based on American national standards.
SCSI interface (Small Computer System Interface)
e ANSI X3.131-1986 An American National Standard)
The connection unit CU on the gateway device GW side on the SCSI interface has a plurality of configurations, and the SC
Access device AU on SI interface and each connection unit
Assign a device identifier ID and a logical unit number LUN to the CU, use a different value for the device identifier ID between the access device AU and each connection unit CU, and assign a different logical unit number LUN between each connection unit CU for access. The information transfer between the device AU and the gateway device GW is performed by distinguishing the transfer destination by the device identifier ID and the logical unit number LUN. Each connection unit CU includes routing processing of the entire gateway device GW in advance, and the gateway device When any of the connection units CU on the GW side has failed, the access device AU replaces the device identifier ID of the failed connection unit CU with a normal device identifier.
A degeneration method in the event of a fault, characterized in that the faulty connection unit CU is cut off by changing to an ID and degenerated to a normal connection unit CU.