KR100348566B1 - Dual storage apparatus in communication system - Google Patents

Abstract

The present invention relates to a redundant storage device of a communication system, comprising: a host monitor for requesting bus use when a disk input / output operation of a host is started and stopping a bus use request when the disk input / output operation is terminated; A bus arbitrator that arbitrates bus usage for a bus monitor of a host monitor operating in an active mode and a bus request of a host monitor operating in a standby mode; A disk access masking unit which prevents access to the disk if the bus is not used in the bus intermediary, and permits access of the disk when the bus is permitted to use the bus; A route setting unit configured to set a data path between a host and a disk when a bus is permitted in the bus intermediator; A synchronization unit for synchronizing disk input / output operations of the active board and the standby board according to the set path; And a disk driver for notifying the start of the disk I / O operation of the host, controlling the progress of the disk I / O operation according to the masking state, and notifying the end of the disk I / O operation according to the synchronization state and whether the disk I / O operation is normally completed. Do.

According to the present invention, the same operation as that of duplication of the memory is performed in hardware for the disk. In addition, even when a large disk such as billing information is required, it can be mounted on one board without powering on a separate board and accessing it through a SCSI controller, thereby saving space and significantly reducing costs.

Description

Dual storage apparatus in communication system

The present invention relates to a storage device of a communication system, and more particularly to a redundant storage device of a communication system.

In general, a communication system includes a plurality of boards for each function, such as a processor board, an input / output processing board, and a data storage board, and each board is connected to each other through a bus. However, the communication system uses a redundant system to ensure reliability, and for this purpose, each board is duplicated into an active board and a standby board. An active board is a board that is currently operating, and a standby board is an active board when an error occurs in the active board.

However, standby boards usually maintain the same contents of the active board's main memory (including the stack). Therefore, if a switchover to the standby board occurs due to a fault in the active board, the standby board can continue the work currently being performed by using the data in its main memory. This is because all data related to the work in progress is located in main memory. System information or task-related data (such as billing information) that are not contained in the main memory are mainly present in nonvolatile storage.

When implementing a nonvolatile storage device in a communication system, a flash memory, a battery backup memory, and a hard disk are used according to each characteristic. In order to ensure the reliability of the system, it is redundant by hardware method such as main memory, but usually the software is used to store data in two media.

Flash memory takes up less space and has faster read times than DRAM, which has disadvantages such as the cost of DRAM, interruption of service during the time to erase blocks, and the limitation of the number of writes. Therefore, it is suitable for applications in which there are no writes and continuous read operations during operation such as storage of an operating system or system programs. When used for this purpose, data redundancy is unnecessary.

And products with RAM with battery backup, such as NVRAM with a battery attached to SRAM, can only be used in extremely limited applications because their maximum capacity is only a few megabytes. The same applies. However, implementing a memory of tens to hundreds of megabytes is too costly and space-consuming and very limited to use.

In addition, the product implemented by back-up with battery in general DRAM has the advantages of reasonable price and fast access time, while the physical size is large, there is a change of existing circuit, and the capacity of battery is variable according to maximum power failure. There is a difficulty.

Hard disks are disadvantageous due to their large size, limited interface, and access time of several milliseconds, but they are suitable as mass storage devices for communication systems in that they can realize large capacity (several GBs) and have significant advantages in terms of price / capacity. . However, due to its size and + 12V supply, it must be implemented as a separate board and accessed through a SCSI controller.

The technical problem to be solved by the present invention is to solve the above-described problems, the same operation as the duplication of the memory to be operated in the hardware to the disk in order to operate in two modes of active / standby (one) It is to provide a redundant storage device of communication system that can be installed in the board of the board to save cost and space.

1 shows a schematic block of an active board and a standby board of a communication system in which a redundant disk is used.

2 shows a schematic appearance of a redundant disk module.

Figure 3 shows a block diagram of an embodiment of a redundant storage device according to the present invention.

In accordance with another aspect of the present invention, a redundant storage device of a communication system includes at least one active board and a standby board which acts as an active board when a defect occurs in the active board, respectively. Includes a removable disk, and the active board and the standby board are connected to each other via a bus. When the disk control block of the active board and the standby board is called a host, a disk input / output operation of the host is performed. A host monitor requesting bus use when started and stopping the bus use request when the disk input / output operation is terminated; A bus arbitrator that arbitrates bus use for a bus request of a host monitor operating in an active mode and a bus request of a host monitor operating in a standby mode; A disk access masking unit for preventing access to a disk if the bus is not used in the bus intermediator, and allowing the disk to be accessed when the bus is permitted; A route setting unit configured to set a data path between a host and a disk when a bus is permitted in the bus intermediator; A synchronization unit for synchronizing disk input / output operations of the active board and the standby board according to a path set through the path setting unit; And notifying the start of the disk I / O operation of the host, controlling the progress of the disk I / O operation according to the masking state, and notifying the end of the disk I / O operation according to the synchronization state of the synchronization unit and whether the disk I / O operation is normally completed. It is preferable to include a disk driver.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. 1 is a schematic block diagram of an active board and a standby board of a communication system in which a redundant disk, which is an embodiment of a redundant storage device according to the present invention, are used. As shown in FIG. 1, the active board and the standby board are normally connected to a backplane board, and a disk of the redundant disk module may be loaded onto the active board and the standby board. Figure 2 shows a schematic appearance of a redundant disk module, consisting of a redundant board with a built-in redundancy logic and a hard disk removable to the redundant board.

3 is a block diagram showing an embodiment of a redundant storage device according to the present invention, which includes a host monitor 310, a bus mediator 340, a disk access masking unit 320, a routing unit 350, and the like. The base 330 and the disk driver 300. In this embodiment, it is assumed that an IDE type disk is used. In case of IDE type disk, this is notified by interrupt when disk I / O operation is completed, and status register is indicated to indicate whether disk I / O progresses (Busy bit) and disk operation is completed normally (done bit, error bit). .

The host monitor 310 requests the bus to be used when the disk I / O operation is started, and stops the bus to be used when the disk I / O operation is finished. In more detail, when disk I / O is started while monitoring the operation of the host IDE bus, the bus arbiter 340 is requested to use the bus, and when the disk I / O ends, the bus use request is stopped. If the redundant storage device (hereinafter, referred to as EDM) according to the present invention operates as a standby, the request of the bus is transmitted to the bus mediator of the active EDM on the Remote IDE bus. When the host monitor 310 operates in the standby mode, that is, when the remote monitor is used as a remote IDE bus monitor, the host monitor 310 determines whether the EDM at the end of the remote IDE bus is Alive. If the host monitor 310 is Alive, the host unit 310 notifies the synchronization unit 330 of a related signal for disk redundancy.

The bus arbitrator 340 arbitrates the bus usage for the bus request of the host monitor operating in the active mode and the bus request of the host monitor operating in the standby mode. For disk redundancy, data must be written to two disks at the same time. To do so, the host IDE bus of Active EDM must occupy its own disk and the disk bus of standby EDM. For this, arbitration logic is required, and as the input of this logic, a signal output from the host monitor 310 of each EDM is used. Implement Fairness to eliminate Starvation. For example, if there is a bus request on the active board and the standby board at the same time, if the bus license is obtained from the host of the active board, the host of the standby board will receive the bus license next.

The disk access masking unit 320 prevents the disk from accessing the bus when the bus mediator 340 does not permit the use of the bus and permits the disk access if the bus usage is permitted. In this embodiment, since an IDE type disk is used, the busy bit of the state register is masked to implement this. In other words, if the host IDE bus initiates disk I / O and issues a command with the disk, it reads the status register and issues a command if the disk is not busy; The host monitor 310 initiates the start of disk I / O and makes a bus use request using the bus. Until the permission is issued, the bus cannot be used until the host monitor 310 starts to use the bus. Mask the bits. When the bus is licensed, it releases masking and proceeds with normal disk I / O.

The routing unit 350 sets a data path between the host and the disk when the bus is allowed to be used by the bus intermediator 340. Interconnect or redirect the three buses, the host IDE bus, the Remote IDE bus, and the disk IDE bus. Here you can configure between host and disk, between host and remote, and between remote and disk.

The synchronizer 330 synchronizes disk input / output operations of the active board and the standby board according to the path set through the path setting unit 350. When working in redundancy, if you write on the host IDE bus of the active EDM, you must also write it to the standby disk. However, if the disks are not synchronized, one disk I / O cannot be terminated equally due to different access time. For this reason, continuous disk I / O may not be possible.

The disk driver 300 notifies the start of the disk I / O operation of the host, controls the progress of the disk I / O operation according to the masking state, and determines whether the synchronization state of the synchronization unit 330 and the disk I / O operation are normally completed. Therefore, it informs the end of disk I / O operation. It is usually embedded in the host's IDE controller (hereinafter referred to as IDEC) and implemented as software.

The operation of the present invention will be described. The storage device in the present invention operates in two main modes. The first is when the host IDEC of the active board performs disk I / O. The second is when the host IDEC of the standby board reads its own disk.

First, a description will be given of a case where disk ID / O is performed by the host IDEC of the active board. First, in the disk driver 300, the host IDEC of the active board predetermines a unique signal indicating the start and end of the disk I / O.

The host monitor 310 of the active board keeps checking whether disk I / O is started. In addition, the IDEC 30 checks the busy bit of the status register. The busy bit of the state register is initially masked as busy. The IDEC 30 continues to poll until the busy bit masking bit is released.

When the host monitor 310 detects the start of the disk I / O, it requests the bus intermediator 340 for the bus.

The bus arbitrator 340 checks the bus request of its active board host and the bus request of the standby board host, and performs fairness arbitration. Therefore, if only the bus request of the host IDEC 30 is present, a bus grant is output. The bus grant goes to the disk access masking unit 320 to perform state register masking, the routing unit (not shown) and the routing unit 350 of the standby board.

The route setting unit 350 sets a path between the host and the disk and between the host and the remote to receive the grant signal.

The disk access masking unit 320 releases the busy bit masking upon receiving the grant signal.

The path setting unit (not shown) on the standby board sets a path so that the disk I / O of the active board is performed on its own disk upon receiving the grant signal.

On the other hand, when the active board Host IDEC 30 confirms that the busy bit of the status register has been released, the active board Host IDEC 30 performs disk I / O operations such as transmitting a command and data according to the set path.

Each disk of the active board and the standby board outputs an interrupt when the disk I / O operation is completed. The interrupt is input to the synchronizer 330. The synchronizer 330 logically ANDs the interrupts to synchronize the disks, and when the I / O operations on both disks are completed, the synchronizer 330 notifies the host IDEC 30 that the disk I / O is finished.

When the host IDEC 30 receives the completion signal of the disk I / O from the synchronizer 330, the host IDEC 30 reads the status register to check the work completion bit of the status register. When the operation is completed, the disk driver 300 performs a unique operation for notifying the end of the disk I / O operation. If the task is not completed, retry or exit immediately. This is a problem to be determined according to the user's convenience.

The host monitor 310 disables the bus request signal REQ requested by the bus intermediator 340 when detecting the end unique operation of the disk I / O. The bus intermediator 340 then completes the task by disabling the grant signal.

Meanwhile, the DISK I / O on the standby board host, which is the second operation mode, will be described. This differs only in a few ways from DISK I / O in the active board host, the first mode of operation, and the others are similar. The difference is that when the host monitor 310 requests a bus, the bus request signal REQ is transmitted to the bus intermediator of the active board. Since the second mode is a read in the standby board, the routing unit 350 Routing in the host is the path between the host and disk in the standby board. Thus, the synchronization unit is bypassed.

Through the above-described configuration and operation, the present invention is compatible with the IDE disk. In other words, when paired, EDM itself enables disk redundancy. Disk write access in active mode writes to both its own disk and the disk on the standby side. In active mode, read access reads from its disk. All access under standby mode is done on its own disk.

The function of the present invention is briefly described as follows. It has transparency. In other words, the disk IO accessed by the host IDE bus is transparently transferred to its own disk.

Redundancy is possible. In other words, when EDM is operating in active mode, it forwards the disk IO of the host IDE boss to its own disk and to the Remote IDE bus. Mirroring is implemented in disk I / O units consisting of several bus cycles.

In mediation, when EDM is running in active mode, the host IDE bus writes to its own disk, but at the same time to the standby disk on the Remote IDE bus. In this case, the standby module and the arbitration process are preceded.

Standby Write is supported. When the EDM is operating in standby mode, it writes its own disk to the redundant write operation that is sent through the RemoteIDE bus.

On the other hand, the present invention has been described by taking an IDE bus as an example by connecting to the host module in the active board and standby board. However, the IDE bus can be replaced with a 68K processor bus, a PCI bus or another bus. That is, if only the bus interface is created and added, other bus interfaces other than the IDE are included in the present invention. Since the bus interface can be easily implemented by those skilled in the art to which the present invention pertains, no separate description will be made herein, and thus the present invention is not limited thereto.

According to the present invention, the same operation as that of duplication of the memory is performed in hardware for the disk. Therefore, it can be operated in two modes, active / standby.

In addition, even if a large capacity disk is required, such as billing information, it can be mounted on a single board without the need to power a separate board and access it through a SCSI controller. The result is space savings and significant cost savings.

Claims (1)

At least one active board and a standby board operating as an active board in the event of a failure of the active board, each of the active board and the standby board includes a removable disk, and the active board and the standby board are connected to each other through a bus. In the communication system to be connected, when the disk control block of the active board and the standby board is called a host, A host monitor requesting the use of a bus when the disk input / output operation of the host is started, and stopping the request of the bus when the disk input / output operation is terminated; A bus arbitrator that arbitrates bus use for a bus request of a host monitor operating in an active mode and a bus request of a host monitor operating in a standby mode; A disk access masking unit for preventing access to a disk if the bus is not used in the bus intermediator, and allowing the disk to be accessed when the bus is permitted; A route setting unit configured to set a data path between a host and a disk when a bus is permitted in the bus intermediator; A synchronization unit for synchronizing disk input / output operations of the active board and the standby board according to a path set through the path setting unit; And A disk for notifying the start of a disk I / O operation of a host, controlling the progress of the disk I / O operation according to the masking state, and notifying the end of the disk I / O operation according to the synchronization state of the synchronization unit and whether the disk I / O operation is normally completed. Redundant storage in a communication system comprising a driver.