JP2001022526A - External storage device and its operating method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable multiple processes to operate and to shorten a process time by composing an external storage device of one main module and multiple extension drive modules and performing the processes on the different extension drive modules. SOLUTION: The external storage device 200 is connected to a host device 100. The external storage device 200 comprises one main module 300 and multiple extension drive modules 401 to 40m, which are connected by fiber channel interfaces. The main module 300 comprises a controller part 310 which controls the whole external storage device 200, single or multiple drives 701 to 70n for data storage (n: natural number one or more), and switches 320 to 32m. The switches 320 to 32m separate or integrate the main module 300 and the extension drive modules 401 to 40m to control a flow of data, indications, and other instructions between both the modules.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an external storage device connected to a computer system, a network system and other electronic information processing organizations, and more particularly to an external storage device having a large number of magnetic disk devices as data storage units. In this case, the present invention relates to a connection configuration of data storage units (so-called back-end connection configuration) suitable for shortening the time of the initial format and a method of multiplexing these processes.

[0002]

2. Description of the Related Art In recent years, there has been an increasing number of opportunities to handle huge data such as weather information and moving images and audio in digital broadcasting of high-definition television. As a result, when the external storage device for storing these data is required to have a large capacity, and when the reliability of the data storage is also required,
A disk array device (RAID: Redundan) that improves processing capacity by distributing data to magnetic disk drive units while also storing redundant data to improve reliability.
t Arrays of Inexpensive Disks) are used.

[0003] Regarding disk arrays, see the paper ("A Case
For Redundant Arrays Of Inexpensive Disks (RAID) ",
David A. Patterson, Garth Gibson, And Randy H. Kat
z, Computer Science Division Department Of Electri
cal Engineering And Computer Sciences, University
Of California Berkeley).

An external storage device such as a disk array device has a large number of data storage units mounted thereon, and an interface is required to connect these units. The maximum number is two for the IDE interface, and seven or fifteen for the SCSI (scuzzy) interface. If the number of data storage units is larger than this, the fiber channel interface will be used.

[0005] Since a fiber channel interface can be expected to have higher transfer capability and expandability than conventional interfaces, products using a fiber channel interface as an interface have been put to practical use. Further, in order to increase or expand the storage capacity of the external storage device, an external storage device that additionally connects an expansion housing having a large number of data storage units (data storage drives) separately has been put to practical use. .

[0006]

A) Before actually operating the disk array device, it is necessary to initialize (format) the medium of the data storage unit (magnetic disk device) mounted on the device. This first problem arises because of the large number of magnetic disk devices and the adoption of a RAID configuration.

As a format, there is a simple format in which only a medium corresponding to a header portion is rewritten without verifying a portion corresponding to a data portion on a medium. The simple format has the advantage of being completed instantly, but has low reliability, and there is a possibility that an error occurs in data written by the user, and cannot be recovered if garbled characters occur.

In order to avoid this, it is necessary to guarantee data at the time of formatting. The format described in the present invention is a format having a data guarantee function.

The format for performing data assurance includes, in response to a format request from a higher-level device, a) writing data as an initial value to a data storage unit in all areas to be formatted; When adopted, the redundant data storage unit writes the redundant data calculated from the data as the initial value to all the areas to be formatted, or c) performs both a) and b). That is. For this reason, in the format of an external storage device (including a disk array device) that requires reliability, a format that is proportional to the size (data storage capacity) of a command to many magnetic disk devices and the format target area is provided. Processing time is required.

[0010] The format for ensuring the reliability is mainly implemented when an external storage device is introduced and when an expansion enclosure having a large number of data storage units is separately added. It is time to do.

First, in the initial stage of introduction of the external storage device, the format of the entire system is required. In other words, all data storage units (magnetic disk units)
Must be collectively formatted, and there is a problem that a formatting time proportional to an increase in storage capacity is consumed. In addition, when a storage medium is added, it is required that the format of the added storage medium can be executed without stopping an external storage device (such as a disk array device) that is already operating.

The format is an operation of writing data (write data) of a certain pattern into a data storage location on a medium. In the disk array device, since a redundant data storage location further exists, it is necessary to write a data pattern of the redundant data storage location.

When calculating and writing this redundant data,
Until formatting is completed, write data is continuously generated between the controller unit in the disk array device and the storage medium. For this reason, while the added storage medium is being formatted, a load is applied to the data transmission line (data bus) between the controller unit and the storage medium, so that access to the remaining storage medium (magnetic disk device) and the entire device are performed. Adversely affect the performance of the device.

In order to solve these technical problems, there is a method of performing formatting using a data storage unit (magnetic disk device) incorporating a dedicated semiconductor chip capable of writing data of several types of patterns. In other words, the controller unit that has received a format instruction or other command from the higher-level device issues a vendor-unique command (a function prepared by the manufacturer of the magnetic disk device) to each magnetic disk device, not a format command. Thus, each magnetic disk device performs formatting by writing data of a certain fixed pattern. With this method, the load imposed on the data bus between the magnetic disk devices can be significantly reduced. In addition, the formatting time can be shortened by formatting a plurality of magnetic disk devices independently or in parallel. However, a dedicated data storage unit (magnetic disk device) is required, and a general-purpose magnetic disk device cannot be used, which increases the manufacturing cost of the external storage device.

Therefore, in the present invention, a general-purpose data storage unit is used in an external storage device (including a disk array device) to which a number of data storage units are connected by using a fiber channel interface. It is an object of the present invention to perform high-speed format processing and reduce a load imposed on a data bus during execution of formatting.

B) As a second problem, FCAL (F
In connection by an ibre Channel Arbitrated Loop), one packet (unit of information) passes through all the control devices and data storage units connected to this loop, so the total number of control devices and data storage units increases. Then, there is a problem that the processing speed for the control target closed by the FCAL is reduced. This problem is solved, for example, as disclosed in U.S. Pat. No. 5,768,551.
It is improved by operating the switch to divide the AL. However, this only improves the processing speed in the loop that did not leave the control device or the higher-level device, and does not disclose the processing in the loop that left the control device or the higher-level device.

Therefore, according to the present invention, a command (for example, a write command) from a host device is mainly stored in an external storage device (including a disk array device) to which a large number of data storage units are connected by using a fiber channel interface. After receiving the command and passing the command to the sub-controller connected thereto, the fiber channel loop including the sub-controller is separated as appropriate,
In the loop separated from the main control device, processing such as calculating redundant data of RAID and performing write processing is performed independently and independently, for example, as post-processing of a write instruction.

[0018]

In one embodiment of the present invention, an external storage device is composed of one main module and a plurality of expansion drive modules (FIG. 1). The main module includes a controller unit that controls the entire external storage device connected to the host device, and a plurality of data storage units (magnetic disk devices). The expansion drive module may be installed in the external storage device from the beginning to configure the entire device, or may be installed later for expansion.

The extended drive module may be composed of a sub-controller unit for controlling the internal control thereof and a plurality of data storage units (magnetic disk devices), or may be composed of only a plurality of data storage units. May be. Further, it may have a user interface (FIG. 5). The main module may have the same configuration as the extension module. Here, the user interface does not mean only an interface circuit but an electronic terminal including a display panel, a service processor, and other information processing terminals.

The extended drive module and the main module are connected via a switch for controlling the flow of data and instructions to and from another path, and form a main loop. Unlike the conventional switch (FIG. 2), the switch (FIG. 3) of the present invention has a function of selecting whether the transmission line connected to the outside is formed into one loop or two loops. are doing.

When the host device does not access the controller of the main module, the switches 321 to 32m are OFF (open) as default and (m is 1
(The above natural number) The main module and the extended drive module are separated. Therefore, the control target of the controller of the main module is only the magnetic disk device inside the main module.

Extended drive module 401 to 40 m
Switches the instruction from the bus (FCAL loop) to switch 3.
When the switch is received from 21 to 32 m, the switches are controlled to an ON state (closed) or an OFF state (open) according to the destination and type of the command. When the extended drive module has a sub-controller section as shown in FIG. 5, a transmission line (FCAL loop) for transmitting data and instructions to the sub-controller section is separated from or integrated with the main module loop. A function may be provided.

The controller unit, which has received the instruction from the host device, determines to which of the extended drive modules the instruction is issued to the magnetic disk device (storage medium), and switches to the sub-controller unit on the extended drive module. Issue instructions via. Then, the instruction issue source is taken over from the controller unit to the sub-controller unit,
Responsible for data processing (format, post-processing of write command, etc.) for the target magnetic disk device according to the contents of the instruction. When the processing is completed, the sub-controller reports the completion of the processing to the controller, and the controller reports the completion to the host device. At this time, since the extended drive module can be closed in the extended drive module to execute processing, it does not affect other extended drive modules at all.

If the magnetic disk devices of the main module 300 and the extended drive module 401 do not impose a heavy load on the controller 310, the external storage device 200 is replaced with an extended drive module having no sub-controller unit as shown in FIG. The switches 320 to 32m may be controlled from the controller unit 310.

If the extended drive module of FIG. 5 is used, the processing can be multiplexed by executing the processing on separate extended drive modules, so that the processing can be performed on all the mounted data storage units (storage media). Is required, the time required for the processing can be reduced. Further, even if a process is being executed on a certain extended drive module, it does not affect other extended drive modules at all, so that the performance of another extended drive module does not deteriorate.

[0026]

FIG. 1 is a diagram schematically showing an embodiment of the present invention. The configuration shown in this embodiment is merely an example, and another configuration, for example, an extended drive module shown in FIG. 5 can be used.

The external storage device 200 is connected to the host device 100. The external storage device 200 includes one main module 300 and a plurality of extended drive modules 40.
1 to 40 m, which are connected by a fiber channel interface. The external storage device 200 includes an extended drive module 40m (m is 1
The configuration before the addition of the above (natural numbers) may be used, or the configuration after the addition may be used.

The main module 300 includes a controller 310 for controlling the entire external storage device 200, and single or plural data storage drives 701 to 70n (n
Is a natural number of 1 or more) and the switch 320 to 32 m. The switches 320 to 32m may be provided in the fiber channel interface, or may be provided as separate switch devices. The data storage drives 701 to 70n are magnetic disk devices or optical disk devices that store and store normal data or redundant data. The controller unit 310 is a host device 100
It has the functions of connection with a data storage drive, temporary data storage when data is transferred, and switch control. The optical disk device referred to here includes all types in which the magnetic disk device is modified using optical means to store and reproduce information.

When the extended drive module shown in FIG. 5 is used, the extended drive modules 401 to 40 m
(FIG. 1) includes sub-controller units 411 to 41m (not shown in FIGS. 5, 412 and thereafter) for controlling the entire extended drive module, and single or plural data storage drives 711 to 7mn (FIG. 5, 721 and later) Are not shown), display panels, electronic terminals and other user interfaces 421 to 42m (FIGS. 5, 422) connected to the sub-controller units 411 to 41m (FIG. 5, 412 and thereafter are not shown).
Hereinafter, not shown).

The switches 320 to 32m separate or integrate the main module 300 and the extended drive modules 401 to 40m, and control the flow of data, instructions and other instructions between the two modules. Normally, it is OFF (both modules are separated) and ON (both paths are connected) as needed. It only needs to be turned on while instructions are being exchanged between the two modules. During a normal operation such as a general instruction, the switches 320 to 32m are turned on,
The process is performed while the switch is ON, and the controller unit 310 directly issues a command to the data storage drives 711 to 7mn to perform the process. However, during a special operation such as a format process or a shift / recovery process of the extended drive module to the power saving mode, the switches 320 to 32m are turned ON once and an instruction (command) from the main module 300 side is issued.
Is transmitted to the sub-controller units 411 to 41m on the side of the extended drive modules 401 to 40m, and is immediately turned off.
.About.40 m are separated (when the extended drive module of FIG. 5 is used).

Other special operations include: a) when self-diagnosis of a data storage drive is performed; b) when data is copied between data storage drives in the same extended drive module; In a configuration in which a spare data storage drive is provided in the expansion drive module, data can be transferred between the data storage drive and the spare data storage drive in the same expansion drive module for data backup and recovery in the event of a failure. For example, copying may be performed.

If the sub-controller unit in the extended drive module has a function of controlling its own input / output, that is, a function of opening and closing the switches 321 to 32m, the sub-controller unit is independent of the host device 100 and the controller unit 310. Then, data processing may be mutually performed between the extended drive modules. As an example of the configuration of the external storage device 200 using the extended drive module in FIG. A distributed configuration can be employed.

In FIG. 1, a controller unit 310 and n data storage drives 701 are provided in a main module 300.
70n and switches 320 to 32m are described, but the main module 300 is only the controller unit 310 and the data storage drive 70
A configuration in which 1 to 70n are not arranged at all is also possible. Also,
User interface 42 such as a panel connected to sub-controllers 411 (FIG. 5) to 41m (not shown)
With instructions from 1 to 42 m, it is possible to perform data processing of its own magnetic disk device and other data storage units asynchronously with instructions from the host device 100. Here, the term "asynchronous" means that two or more processes are independently independent of each other with no temporal relationship except that they have a common starting point.

At this time, the execution state and progress of the instruction can be confirmed through the user interface 421 and the like. The user interface 421 and the like are effective when the extended drive module is disposed in a housing different from the main module 300. When it is not necessary to regularly check the execution state or progress of an instruction (instruction), user interfaces 421 to 42m to which a portable electronic terminal can be connected are arranged instead of a fixed display panel. good. When accessing the extended drive module using the portable electronic terminal, the portable electronic terminal may have some or all of the functions of the sub-controller unit 411 in FIG. Also, user interfaces 421 to 4
A configuration in which the 2m is not disposed or a configuration in which the sub-controller unit is not disposed only by the user interface is also possible.

FIGS. 2 and 3 are diagrams schematically showing the operation of the switch used in the present invention and the bypass switch conventionally used. Although the transmission line is represented by one line, it means two lines in a twisted pair, a coaxial cable, a flat cable, and another set of transmission lines.

In the conventional bypass switch (FIG. 2), when the switch is ON, 901 is opened, and 902 and 903 are opened.
Is closed to form a main loop composed of A → E → C → D → F → B → A. 9 when the switch is turned off
01 closes, 902 and 903 open, A → E → F → B
→ A main loop composed of A is formed. That is,
There is only one loop.

On the other hand, the switch used in the present invention (FIG. 3)
911 and 914 open when the switch is ON,
912 and 913 are closed and A → E → G → C → D → H → F
A main loop composed of → B → A is constituted. When the switch is turned off, 911 and 914 are closed and 912 and 913 are opened, and a main loop composed of A → E → F → B → A and a sub loop composed of G → C → D → H → G are configured. Is done. That is, the sub-loop separates from the main loop to form a loop. In general terms, one closed loop composed of a transmission line is defined as two smaller closed loops.

The switch used in the present invention is normally OFF, and it is necessary to exchange commands and data between two modules (regardless of whether the module is a main module or an extended drive module). Only functions and turns ON.

Here, a method of executing the format processing according to the present invention will be described.

For example, as shown in FIG. 1, assuming that the data storage drive 711 is a storage medium to be formatted, the format instruction from the higher-level device 100 is:
The controller unit 310 determines that the instruction is for the data storage drive 711 belonging to the extended drive module 401. The controller unit 310 controls the sub-controller unit 411 that controls the entire extended drive module 401 or the data storage drive 71 by itself.
Issue an instruction to format 1, and wait for the report of formatting completion. At this time, since an instruction transmission request is generated between the main module 300 and the extended drive module 401, the switch 321 reacts and turns ON once, and immediately after transmitting the instruction from the controller unit 310 to the sub-controller unit 411, May be turned off (when the extended drive module of FIG. 5 is used). The switch 322 does not respond to this instruction and
It remains as FF. By using the function of turning off the switch 321 and separating the extended drive module 401 from the main loop, a sub-loop can be formed independently, and the sub-controller 411 (FIG. 5) takes charge of the control.

Then, the sub-controller unit 411 starts formatting the data storage drive 711.
Formatting is the work of writing write data of a fixed pattern according to the capacity of the storage medium to be formatted on the storage medium. At this point, the sub-controller unit 411
And between the data storage drive 711 and the write data I / O by format until the format is completed.
Issued continuously. When the formatting of the data storage drive 711 is completed, the sub-controller unit 411
Reports the completion of formatting to the controller unit 310 when there is no other data storage drive to be formatted. The operation of this processing is the reverse of the procedure for issuing a format instruction, and the sub-controller unit 411 issues a report,
The switch 321 responds and transmits the signal to the main module 300 or directly receives the report by the controller unit 310. The progress of the formatting operation of the data storage drive 711 is indicated by the sub-controller unit 4.
Using the user interface 421 connected directly or indirectly to the host device 11, it is possible to confirm asynchronously with an instruction from the host device 100.

It is also possible to issue an instruction to format the data storage drive 711 using the user interface 421 instead of the host device 100. In this case, the user interface 421 instructs the sub-controller unit 411 to format the data storage drive 711, and the sub-controller unit 41
1 starts formatting of the data storage drive 711. The user interface 421 may directly format the data storage drive 711. The switch 321 does not generate a request to transmit an instruction between modules, and remains OFF, and the extended drive module 401 is separated from the main loop. Subsequent operations are the same as when the format command is received from the higher-level device 100 described above, but a report of the completion of the format is reported not to the controller unit 310 but to the user interface 421. Also in this case, the progress of the formatting operation of the data storage drive 711 can be confirmed using the user interface 421.

When the format processing is executed in this manner, when the transfer of the write data by the format is started between the sub-controller unit 411 and the data storage drive 711, the data transfer in the main module 300 is completely stopped. This becomes unnecessary, and only data transfer by a format operation closed in the extended drive module 401 is performed. At this time, if another instruction is issued from the higher-level device 100 and the instruction is for the data storage drive 712 in the extended drive module 401, the formatting operation is executed in the extended drive module 401. As a result, the performance of the instruction to the data storage drive 712 is lower than normal, due to this influence.
Data storage drives 701 to 70n and the data storage drive 72 in another extended drive module 402
1 to 72n, the data storage drive 7
No load is imposed by the 11 format, and no performance degradation occurs. Further, by performing the formatting process in different expansion drive modules in parallel, even when all storage media need to be formatted, regardless of the number n of expansion drive modules constituting the external storage device 200, Formatting of all extended drive modules can be completed in the same time as formatting performed by one extended drive module.

Here, the external storage device 2 constructed using the bypass switch 331 of the prior art as shown in FIG.
Considering that the format of the data storage drive 711 is executed at 00, the switch is the bypass switch 331 in the related art, so that a new loop cannot be formed by the extended drive module 401 separated from the main loop. Not available. Of course, since the sub-controller unit 411 does not exist, the extended drive module 401 cannot be controlled. Therefore, at the time of formatting the data storage drive 711, the controller unit 310 starts sending write data directly to the data storage drive 711. Therefore, during the execution of the format, both the main module 300 and the extended drive module 401 are in a load state.

[0045]

According to the present invention, when processing is required simultaneously by a plurality of expansion drive modules, the processing can be performed by closing the processing within each expansion drive module.
Multiple processing can be performed, and the time required for the processing can be reduced. In addition, since each extended drive module can perform asynchronous processing without affecting other extended drive modules, even if a process with a large load is performed by a certain extended drive module, the entire external storage device is Has no effect, and normal operation can be guaranteed with other extended drive modules.

Although the processing of the format has been described as an embodiment of the present invention, by adopting the connection configuration as in the present invention,
For example, a self-diagnosis of a data storage drive can be performed by an extended drive module online, data can be copied between data storage drives in the same extended drive module, and data in a plurality of extended drive modules can be copied. Even when transferring data between storage drives or putting one expansion drive module into power saving mode, it can be performed without affecting other unrelated expansion drive modules, etc., and it can be multiplexed Becomes

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram showing one embodiment of the present invention.

FIG. 2 is a diagram illustrating a conventionally used bypass switch.

FIG. 3 is a diagram showing a switch used in the present invention.

FIG. 4 is a diagram for explaining a method of performing a format process in the related art.

FIG. 5 is a diagram for explaining an extended drive module including a sub-controller unit and a user interface.

[Explanation of symbols]

100: Host device, 200: External storage device, 300: Main module, 310: Controller unit, 320 to 32m: Switch, 330 to 33
m: bypass switch, 401 to 40 m: extended drive module, 411 to 41 m: sub-controller unit, 4
21 to 42 m: user interface, 701 to 7 mn
.., Data storage drives, 901 to 903, internal switches in bypass switches, 911 to 914, internal switches in switches.

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Atsushi Ishikawa 2880 Kozu, Odawara-shi, Kanagawa F-term in Storage Systems Division, Hitachi, Ltd. (Reference) 5B014 HC13 5B061 AA00 FF07 5B065 BA01 BA03 CA11 CA30 CC08 CE22 ZA13

Claims (8)

[Claims] 1. A switch having a function of converting one closed loop composed of a transmission line into two smaller closed loops, and a controller connecting two or more switches so as to form one first loop-shaped transmission line. And a module that connects to a higher-level device; a module that connects a plurality of data storage units to one second loop transmission line; and an external memory that connects two or more modules via the switch. apparatus. 2. The external storage device according to claim 1, wherein said module further has a function of connecting an interface circuit to said second loop-shaped transmission line, and controlling said switch via said interface circuit. An external storage device having: 3. The external storage device according to claim 2, further comprising a user interface connected to said interface circuit, wherein said controller issues an instruction for a module having said user interface issued from said higher-level device. When the unit receives the command, the command is issued from the controller unit to the user interface, and the external storage device is executed by the user interface. 4. The external storage device according to claim 3, wherein said instruction is a format instruction for said data storage unit. 5. The external storage device according to claim 2, further comprising a user interface connected to said interface circuit, wherein said controller issues an instruction issued from said higher-level device to a module having said user interface. An external storage device in which the switch functions so as to separate the module having the user interface from the first loop transmission line after the controller issues the command to the user interface; . 6. The external storage device according to claim 3, wherein a progress status of said command is output to said user interface. 7. The external storage device according to claim 3, wherein a progress status of said instruction can be output from said higher-level device. 8. A switch having a function of converting one closed loop composed of a transmission line into two smaller closed loops, and a controller for connecting two or more switches so as to form one first loop-shaped transmission line. And a module that connects to a higher-level device, a module in which a plurality of data storage units and a user interface are connected to one second loop transmission line, and two or more of the modules connected through the switch. A method of operating a connected external storage device, comprising: a first step of issuing a command to the module having the user interface from the higher-level device; a second step of receiving the command of the first step by the controller unit; The command received by the controller unit in the second step is:
A third step of issuing to the user interface; a fourth step of operating the switch connected to a module having the user interface; and a fifth step of executing an instruction received by the user interface.
And an operation method of the external storage device.