JP4378091B2 - Motherboard duplication and disk multiplexing system and RAID control unit in computer apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention is duplicated and a multiplexing technique disks motherboard of cold standby by computer apparatus when combined with RAID system.
[0002]
[Prior art]
As a host (motherboard) duplication technology, the duplex method of duplicating the motherboard and the disk while originally executing the business, and executing the original business on both the mother board and the disk pair to match the execution results of both There is a dual method that advances business execution while constantly checking.
In addition, in the duplex system, a cold standby that duplicates the motherboard and uses one for the original business execution, switches the other motherboard to a non-energized state and switches to the other motherboard to perform the original business when a failure occurs. There is a hot standby system in which a main system executes an original business (for example, transaction processing) and a sub system executes another business (for example, batch processing).
[0003]
In the cold standby system, as shown in FIG. 8 (a), when a failure occurs, a mother board (82, 82 ') and a disk (83, 83') are duplicated and one of them is used for the actual execution of business. Switch to the other and execute the business originally, and as shown in FIG. 8B, only the motherboard (82, 82 ') is duplicated, and the disk 83 is shared by the motherboards 82, 82' and one motherboard is used. There is a method that is originally used to execute a business and switches to the other when a failure occurs to execute the original business.
[0004]
The motherboard is usually a substrate that forms the center of the host computer and is a substrate (board) in which basic components of the host computer such as a CPU, a RAM, and various interfaces are incorporated.
[0005]
However, when the host (motherboard) is duplicated by the cold standby method and combined with the RAID1 system (see Non-Patent Document 1), two hosts are connected on the host side as shown in FIG. Since the power is supplied only to the power supply, the non-powered motherboard must be electrically disconnected. Further, there is a technical problem of how to switch the connection with the disk device.
[0006]
Here, as a conventional technique, in a disk array system, a unidirectional signal driver that transmits signals from the A side only to the B side in order to accurately relay bidirectional interface signals without increasing the number of interface signals. And a bidirectional signal driving unit that performs bidirectional signal transmission from the A side to the B side and from the B side to the A side. The bidirectional driving unit includes an output buffer unit that outputs bidirectional signals, and a bidirectional signal driving unit. There is an interface relay device of a disk array system that includes a bidirectional signal output section that sends the above signal as an input to an output buffer section (see, for example, Patent Document 1).
[0007]
In addition, when a RAID control device fails, in order to prevent the RAID control system and the entire system including the RAID control system from being stopped when replacing the failed RAID control device, each of the two RAID control boards is provided. A SCSI (Small Computer System Interface) bus switching unit is provided, and the lower RAID control board selected as the operation RAID control board includes a plurality of SCSI control units that receive control data from the central control unit by the SCSI bus switching unit. In the upper RAID control board that is connected to the HDD device and not selected as the operation RAID control board, the SCSI bus switching unit is configured to disconnect the SCSI control unit from the plurality of HDD devices, and the lower RAID control board When is broken There is a RAID control system in which a SCSI bus switching unit of a higher-level RAID control board is switched and the SCSI control unit is connected to a plurality of HDD devices (see, for example, Patent Document 2).
[0008]
In addition, since the storage device connected to the interface bus is duplicated without mounting a RAID controller, the magnetic disk control device connected to the host computer via the interface bus and the magnetic disk device comprising the magnetic disk Redundant controller, bus monitor, magnetic disk control device, backup magnetic disk, and backup magnetic disk device having changeover switch connected to interface bus in addition to configuration, and storage device duplexing method and storage device There is a failure recovery method by duplication (see, for example, Patent Document 3).
[0009]
[Patent Document 1]
JP 2002-8276A [Patent Document 2]
JP-A-9-319526 [Patent Document 3]
JP 2001-337788 A [Non-Patent Document 1]
Patterson, DA, Gibson, G, Katz, RH, "A Case for Redundant Arrays of Inexpensive Disks (RAID)," Report No. UCB / CSD 87/391, Computer Science Division, University of California Berkeley 1987
[0010]
[Problems to be solved by the invention]
According to the interface relay device of the disk array system disclosed in Patent Document 1, bidirectional interface signals can be accurately relayed without increasing the number of interface signals, but it is not suitable for host duplexing. There was a problem.
[0011]
Further, according to the SCSI control system disclosed in Patent Document 2, when a RAID control device fails, the RAID control system and the entire system including the RAID control system are stopped when the failed RAID control device is replaced. However, there is also a problem that it is not suitable for host duplexing.
[0012]
The method disclosed in Patent Document 3 is a technique related to duplication of storage devices, and is suitable for an interface using SCSI, and is adopted by most hard disks and CD-ROMs of personal computers currently on the market. There is a problem that it cannot be applied to a low-cost IDE (Integrated Drive Electronics) interface, and it is also not suitable for host duplication.
[0013]
The present invention relates to the duplication of a mother board in a computer device that enables electrical disconnection of a non-energized mother board and switching of connection with a disk device, which are necessary when a host (mother board) duplication by a cold standby and a RAID system are combined. An object is to provide a disk multiplexing system and a RAID control unit.
[0014]
[Means for Solving the Problems]
In order to solve the above problems, in the first aspect of the present invention, the first and second motherboards provided in a single computer device, a plurality of disk devices constituting the disk sharing arrangement for connecting the computer apparatus In between, there is provided one control unit for controlling the access operation to the plurality of disk devices by energizing one of the first and second motherboards and not energizing the other motherboard. There is provided a system for duplicating a mother board and a disk device in a computer device .
Thereby, it is possible to easily realize the duplexing of the mother board and the multiplexing of the disk device by the cold standby with one control unit .
[0015]
Further, in the invention of claim 2, wherein, the plurality of disk devices without a RAID1 configuration, write data simultaneously to all of the disk device at the time of writing, data from a predetermined one disk device of the plurality of disk devices in read 2. A system for duplicating a motherboard and a multiplexing system for a disk device in a computer apparatus according to claim 1, wherein:
Thus, in one control unit can be realized multiplexing of disk device in duplication and RAID1 motherboard in the computer device easily.
[0016]
According to a third aspect of the present invention, there is provided a RAID control unit that is connected to the first and second motherboards provided in one computer device and is connected to a plurality of disk devices ,
First and second host interfaces connected to the first and second motherboards via a bus, a plurality of device interfaces connected to the plurality of disk devices on a one-to-one basis via a bus, and a single controller that performs an automatic transfer control registers control of the first and second host interface data via the interface,
In operation, the first and second one of the motherboard and via the host interface and the plurality of device interfaces for that motherboard perform automatic data transfer control between the plurality of disk devices, but other one There is provided a RAID control unit characterized in that it does not output a host interface corresponding to a motherboard of the above and does not accept an input signal.
Thereby, it is possible to easily realize the duplexing of the motherboard and the multiplexing of the disks in the RAID 1 in one computer. In addition, the electrical risk at the time of disk replacement by hot swap can be minimized.
[0018]
According to a fourth aspect of the present invention, there is provided the RAID control unit according to the third aspect, wherein the host interface and the device interface are IDE interfaces.
As a result, it is not an expensive SCSI or FC (Fibre Channel) that has been used when a disk device is shared by a plurality of hosts as in the past, but an inexpensive IDE that can only be used in a one-to-one connection in the conventional technology. Multiplexed disk devices connected to the interface can be realized.
[0019]
Further, in the fifth aspect of the present invention to provide a RAID control unit according to claim 3 or 4, characterized in that a bidirectional buffer between each device interface and the controller. This configuration makes it possible to separate a failed disk. For example, in the case of a two-disk configuration, one unit can be operated, and in the case of a three-disk configuration, two units can be operated (so-called degenerate operation). It becomes. Similarly, with this configuration, the power-on time of the disk can be completely shifted between two (or more) disks, thus supporting the rapid capacity increase of the disk in recent years (and hence the inrush current at startup). It is also easy to do.
[0020]
Further, in the invention described in claim 6, a RAID control unit that connects the three or more disk devices while connected to the motherboard via a IDE interface for a host connected to the motherboard, a plurality of disks and bus A plurality of device IDE interfaces connected in a one-to-one relationship, one controller that performs automatic data transfer control and register control of the host IDE interface via each device IDE interface , and each device IDE interface And a driver element respectively provided between the controller and the controller, and during operation, automatic data transfer control with a plurality of disk devices is performed via the motherboard host IDE interface and each device IDE interface. R To provide the ID control unit.
As a result, the expensive SCSI and FC that have been used when a disk is shared by a plurality of hosts as in the prior art, but not the expensive IDE interface that can only be used in a one-to-one connection in the prior art. Multiplexing of disk devices connected by more than one unit can be realized.
In addition, since a bi-directional buffer is provided between the IDE interface for each device and the controller , this configuration makes it possible to separate a failed disk. In the case of this disk configuration, operation with one unit (so-called degenerate operation) is possible. Similarly, with this configuration, the power-on time of the disk can be shifted between three (or more) disks, thus supporting the recent rapid capacity increase of the disk (and hence the inrush current at the time of startup). Will also be easier.
[0022]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is an explanatory diagram of a RAID host duplexing and disk multiplexing method according to the present invention. From the power supply 11, motherboards 12 and 13, and disks 15 and buses 5 and 6 within the broken line in FIG. This configuration is the cold standby disk sharing configuration of FIG. In the RAID system of the present invention, the host duplexing and disk multiplexing (hereinafter referred to as "duplexing system") configuration is the configuration described later (FIG. 3) between the motherboards 12 and 13 and the disk 15 in the configuration of this disk sharing system. The RAID control unit 14 provided is provided, connected to the motherboards 12 and 13 via the buses 5 and 6, and added with disks 16 and so on, thereby enabling multiplexing of disks in RAID1.
[0023]
FIG. 2 is a block diagram showing an embodiment of a host duplexing and disk multiplexing system (hereinafter referred to as “duplexing system”) according to the present invention.
In FIG. 2, a “redundant system” including a power source 11, a motherboard (MB 1) 12, (MB 2) 13, a disk 15, buses 5 and 6, a RAID control unit 14, and buses 7-1 and 7-2. 1 shows a host computer 10 including a main part of FIG.
[0024]
As will be described later (FIG. 3), the RAID control unit 14 is physically connected to both the mother boards 12 and 13, but is electrically connected to only one of them (that is, energized). In this example, two disks (HDD1) 15 and (HDD2) 16 are connected. The disks (HDD1) 15 and 16 have a mirror configuration (RAID1).
[0025]
In this embodiment, the RAID control unit 14 can connect two or three disks, but may be three or more. The disk is not limited to a hard disk, and may be a storage device that can access a storage medium such as a CD-ROM.
<RAID control unit>
As shown in FIG. 3, the RAID control unit 14 includes a microcomputer interface and five IDE interfaces (host interface × 2, disk interface × 3), and performs data transfer between the host (motherboard) and the disk in accordance with instructions from the microcomputer. . Three disks can be operated simultaneously, but only one of the motherboards. Note that the output of the host interface that has not been selected is High-Z, and no input signal is accepted (this enables hot swapping).
[0026]
IDE (Integrated Drive Electronics) is an interface employed by most hard disks and CD-ROMs at the present time, and has a simple structure and does not require an adapter card for control, so it is cheaper than SCSI. However, the host interface is not limited to IDE.
[0027]
FIG. 3 is a block diagram showing a configuration example of the RAID control unit, and shows an example in which three disks are connected.
Specifically, as shown in FIG. 3, the RAID control unit 14 is connected to a host (motherboard (MB)) duplex with an IDE interface (host interface (HOST I / F)) having two ports (HOST1 I / F, HOST2 I / F) is provided to allow access from both motherboards (MB1, MB2), the command received from the motherboard 12 or 13 is analyzed by the microcomputer 141, and data is transferred by controlling the controller 145. However, simultaneous access is prohibited.
[0028]
3, the RAID control unit 14 includes a one-chip microcomputer 141 that controls the entire board of the RAID control unit 14, a RAM 142, a flash memory (Flash) 143, a battery 144 that backs up the RAM 142, an automatic data transfer control, and a host. A controller (FPGA) 145 for register control of the interfaces 146 and 147, an IDE interface (host interfaces 1 and 2 (HOST1 I / F, HOST2 I / F)) 146 and 147, error notification to the user and reading of a log IDE interface (device interface (HDD1I / F, HDD 2 / F, HDD3 I / F)) that performs serial access (RS232C) 148 and individual access and write processing to disks 15, 16, and 17 in parallel (simultaneously) 151, 161, 171 I have.
[0029]
In the embodiment, SARAM is used as the RAM 142, but the present invention is not limited to this. In addition to being used as a work memory for the microcomputer 141, the RAM 142 stores error information and recovery information. However, since important log data is backed up in the flash memory 143, the operation is not affected even if the battery capacity is exhausted. The flash memory 143 stores the control program of the microcomputer 141 and the data of the controller 145. A serial interface (RS232C) 148 is provided for communication with the host. As communication information, error reports, error statuses are provided. System log data, comparison execution instructions, read master switching instructions, and the like.
[0030]
When two disks are connected, the IDE interface 171 may be left idle or the IDE interface 171 may not be provided.
[0031]
<Hot swap operation>
When exchanging an error disk (hot swap) during operation, the FET (Field-Effect Transistor) is controlled softly by a key switch to shut off the power supplied to the disk. This operation is performed by polling the key switch with the microcomputer 141, and by turning off the power supply to the disk and setting all the disk interfaces in the controller 145 to the High-Z state, the electrical risk at the time of disk replacement is minimized. suppress. In addition, when a newly inserted disk is turned on, it is turned on gently without turning on the power suddenly so as not to place a burden on the operating disk.
[0032]
<Operation example of RAID control unit>
FIG. 4 is an explanatory diagram of an operation example of the RAID control unit at the time of write, read, or rebuild. FIG. 4 (a) is a write operation, FIG. 4 (b) is a read operation, and FIG. 4 (c) is a rebuild operation. It is explanatory drawing of.
The data transfer control in the description of FIGS. 4A, 4B, and 4C is the controller 145 of the RAID control unit 14 in the case of UDMA (short for Ultra DMA / 33, ATA / IDE extended specifications) transfer. Although it is performed in hardware in FIG. 3), it is not limited to this. Further, it is not limited to UDMA transfer (for example, PIO (Programmable I / O) transfer may be used). Data transfer is performed between the host (motherboard) and the disk, or between the disk and the disk. In this case, the transfer sector number, the transfer direction, and the transfer mode are completed when the number of stayed sectors × 256 words is transferred according to the instruction from the microcomputer 141. In the example of FIG. 4, the mother board 13 is not electrically connected.
[0033]
(Example of write operation)
In FIG. 4A, when there are three disks (disks 15, 16, and 17), the mother board 12 writes three simultaneously, and when there are two disks, two simultaneously write. In other words, after issuing a command to each device (disk in this example), writing is performed simultaneously after all devices are ready (REDY).
[0034]
(Read operation example)
In FIG. 4B, the motherboard 12 reads from the read master disk regardless of whether there are three disks (disks 15, 16, 17) or two disks. The read master disk is a primary disk (the uppermost disk 15) by default, but can be changed intentionally from the host side in addition to automatic change after rebuilding.
[0035]
(Rebuild operation)
In FIG. 4C, if the disk after the error disk is replaced by hot swap is the disk 16 ′, the rebuild operation is automatically performed. This operation copies normal disk 15 data to the replaced target disk (= disk 16 ') and does not depend on the file system or partition. However, if the target disk has a smaller capacity than the normal disk, the partition information of the normal disk is confirmed, and if the copy is possible, the rebuild operation is performed. For example, when the rebuild operation is performed, the system can be performed without stopping. Even if the power is accidentally turned off during the rebuild operation, the rebuild operation can be continued at the next startup.
[0036]
<Modification 1>
FIG. 5 is an explanatory diagram of a modification of the RAID control unit. FIG. 5A shows an example in which two disks are connected, and FIG. 5B shows an example in which three disks are connected.
When a disk fails, the bus may be locked (for example, shorted to the adjacent wiring or ground). In the present invention, a driver element (bidirectional buffer: for example, between the buses 7-1, 7-2 (, 7-3) connecting the RAID control unit 14 and the disks 15, 16 (, 17) as shown in FIG. (TTL245 series) 51, 52 (, 53) are provided. In this way, the failed disk can be disconnected from the driver interface (151, 161 (, 171) (FIG. 3)), and in the case of a two-disk configuration, the operation is performed by one unit and the three disks. In the case of the configuration, operation with two units (so-called degenerate operation) is possible. Similarly, with this configuration, the power-on time of the disk can be completely shifted between two (or more) disks, thus supporting the rapid capacity increase of the disk in recent years (and hence the inrush current at startup). It is also easy to do.
[0037]
(Modification 2)
FIG. 6 is a diagram showing an embodiment of disk multiplexing using an IDE interface, and FIG. 7 is an explanatory diagram of a conventional example of disk multiplexing using an IDE interface.
[0038]
Normally, only two devices can be connected to one channel in an IDE interface. Conventionally, only two disks can be connected to one host as shown in FIG. 7, and 3 using an inexpensive IDE interface. The disk array device could not be multiplexed with more than one disk.
[0039]
On the other hand, since SCSI and FC do not have such restrictions, conventionally, expensive SCSI and FC have been used for three or more disk array systems. However, as shown in FIG. 6, driver elements (both of them) are connected between buses 7-1, 7-2, and 7-3 connecting the RAID control unit 14 (FIG. 3) having an IDE interface and the disks 15, 16, and 17. Direction buffers 51, 52, 53). By doing so, the disk array system can be multiplexed (three or more) using an inexpensive IDE interface. Further, by providing the driver element, the failed disk can be disconnected from the driver interface (151, 161, 171 (FIG. 3)) as in the first modification described above, and so-called degenerate operation is possible. Similarly, with this configuration, the power-on time of the disk can be completely shifted between three (or more) disks, thus supporting the recent capacity increase of the disk (and hence the inrush current at startup). It is also easy to do.
[0040]
In the above description, the number of disks that can be connected to the RAID control unit 14 is two or three. However, the present invention is not limited to this, and three or more disks may be connected. In the embodiment, the RAID system is RAID1, but the scope of application of the present invention is not limited to RAID1.
[0041]
As mentioned above, although one Example of this invention was described, this invention is not limited to the said Example, It cannot be overemphasized that various deformation | transformation implementation is possible.
[0042]
【The invention's effect】
As described above, according to the first to fifth aspects of the present invention, duplexing of hosts and multiplexing of disks can be easily realized.
Further, according to the invention described in claim 6, multiplexing of disks (three or more) can be easily realized.
[Brief description of the drawings]
FIG. 1 is an explanatory diagram of a method for duplexing hosts and multiplexing disks in a RAID system according to the present invention.
FIG. 2 is a block diagram showing an embodiment of a host duplexing and disk multiplexing system according to the present invention.
FIG. 3 is a block diagram illustrating a configuration example of a RAID control unit.
FIG. 4 is an explanatory diagram of an operation example of a RAID control unit at the time of writing, reading, or rebuilding.
FIG. 5 is an explanatory diagram of a modified example of the RAID control unit.
FIG. 6 is a diagram showing an embodiment of disk multiplexing using an IDE interface.
FIG. 7 is an explanatory diagram of a conventional example of disk multiplexing using an IDE interface.
FIG. 8 is an explanatory diagram of a cold standby method.
[Explanation of symbols]
5, 6 Bus 11 Power supply 12, 13 Motherboard (host)
14 RAID control unit 15, 16, 17 disk (disk array device)
51, 52, 53 Driver element (bidirectional buffer)
145 Controller 146, 147 Host interface 148 Serial interface 151, 161, 171 Device interface

Claims (6)

A first and second motherboards provided in a single computer device, between a plurality of disk devices constituting the disk sharing arrangement to be connected to the computing device, one of the first and second motherboards Duplicating a motherboard in a computer device , wherein a control unit is provided for energizing one side to control an access operation with the plurality of disk devices and controlling not to energize the other motherboard. A disk unit multiplexing system. Wherein the plurality of disk devices without a RAID1 configuration, characterized in that, for reading data from a predetermined one disk device of all of the write data simultaneously to the disk device, the plurality of disk devices in read during write 2. A system for duplicating a mother board and a disk device in a computer apparatus according to claim 1 . A RAID control unit for connecting to a first and second motherboard provided in one computer device and for connecting to a plurality of disk devices ,
First and second host interfaces connected to the first and second motherboards via a bus, a plurality of device interfaces connected to the plurality of disk devices on a one-to-one basis via a bus, and a single controller that performs an automatic transfer control registers control of the first and second host interface data via the interface,
In operation, the first and second one of the motherboard and via the host interface and the plurality of device interfaces for that motherboard perform automatic data transfer control between the plurality of disk devices, but other one The RAID control unit is characterized in that it does not output the host interface corresponding to the motherboard and accepts no input signal. 4. The RAID control unit according to claim 3, wherein the host interface and the device interface are IDE interfaces. RAID control unit according to claim 3 or 4, characterized in that a bidirectional buffer between the said respective device interface controller. A RAID control unit that connects the three or more disk devices while connected to the motherboard,
A host IDE interface connected to the motherboard; a plurality of device IDE interfaces connected one-to-one via the buses to the plurality of disks;
One controller for performing automatic data transfer control via the device IDE interface and register control of the host IDE interface;
A driver element provided between the IDE interface for each device and the controller;
With
In operation, the RAID control unit performs automatic data transfer control with the plurality of disk devices via the host IDE interface and the device IDE interface of the motherboard.

Images