JPH08272666A - Data backup method, mirroring device, and its control method - Google Patents

Abstract

PURPOSE: To surely back up data by simultaneously performing the distributed preservation operation of the same data toward plural safer separated places. CONSTITUTION: When data generated in A1 which uses a disk HDD1 in a server S1 as a shared disk is written in the disk HDD1, this data is only written in the disk HDD1 in the conventional system; but in this invention, data is not only written in the disk HDD1 but also transferred to the server S1, and data is written in RAID-A to make it dual on the server S1 side. Further, the server S1 transfers data to a server S2 and/or a server S3 through transmission lines L1 and L3, and the server S2 receives it to write it in RAID-B, and the server S3 receives it to write it in RAID-C. Thus, dual data are secured in RAIDs of servers S1 to S3.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data backup method which can be used in a networked computer system, a hardware which makes it more effective, and a control method thereof.

[0002]

2. Description of the Related Art A conventional backup method in a computer network system is mainly intended to prevent a data loss accident that may occur due to a sudden failure of hardware used in the computer system. There was a problem that a data loss accident might occur. Hereinafter, in what kind of case this may occur will be described with a specific example. At present, many networked computer systems are used in various fields, and with the improvement of the performance, the storage devices used in the systems are also being increased in speed and capacity. The hard disk drive "hereinafter HDD" is currently the mainstream in this storage device, but there is a problem in using this hard disk drive. This is called a crash, and if the HDD causes this crash, HD
It is impossible to read out all or some of the stored data in D or in some cases. When such a state occurs in a computer system used for business use, a large amount of data is lost in an instant, resulting in great damage in terms of time and money. In a computer system used for business use, it is general to use a plurality of computers that are networked, and when the HDD used by the main computer in such a system crashes. Since the entire system may become unusable in some cases, when HDD is used, some means for preventing data loss is usually used.
This is a method of making a backup of the D data. In this backup method, as a method often used in a main computer of a networked computer system for business use. 1. How to do disk mirroring. a. Two partitions are performed on one HDD by software, or the same data is written to each of the two HDDs to duplicate the data. b. 2 H using dedicated mirroring hardware
The same data is written in each DD to duplicate the data. 2. How to do disk duplexing. c. The two HDDs used for one disk mirroring are separate, including the disk channel, and the same data is written to each to duplicate the data to improve reliability. In this case as well, there is a dedicated duplexing mirroring device. 3. RAID (Redundant Array of)
Inexpensi-ve Disks). d. A redundant disk array device that uses HDDs connected in parallel has a high reliability because it has a function for recovering lost data. Are used. These are described as the methods often used in the main computer. In the worst case, if the above c or d is used in the main computer, all the data will be lost at least due to the HDD crash. Is due to what is believed to be preventable. However, even in a computer system using the above method and taking measures against, for example, power interruption or power failure other than the HDD, there is a possibility that all the data may be lost by the conventional backup method. Certainly, if the HDD of the main computer is set to the disk duplexing method, the same data is always written in two separate HDDs, so even if one crashes, the other H
Normal data always remains in the DD, and even when RAID is used, normal data remains due to the data recovery function even if a part of the disk array crashes. But,
The state that this data always remains can be secured when normal natural failure is the cause. In the case of disk duplexing, two HDDs that use separate interface boards and controllers of 2 channels. The failure probability is very low at the same time. Similarly, even if RAID is used, the probability that multiple arrays will fail at the same time that data cannot be recovered is also very low. It is a method that is made up of. However, if only electrical and mechanical natural failures are considered, the probability of this is very low, but if factors other than natural failure are added to this probability, the probability will increase by orders of magnitude. this,
Among the factors other than natural breakdown, the disasters such as earthquakes, fires, and water damage are thought to be the most problematic. If such a disaster occurs, the storage device itself such as HDD or disk array will drop from the installation site in the event of an earthquake,
If all of them are broken or damaged at the same time and cannot be read,
It is conceivable that the equipment around the installation location, walls, or the ceiling may fall, which may damage the entire storage device. Even in the case of a fire, the storage device itself may burn out and become unusable, or the storage device may become unusable due to an increase in ambient temperature, or the storage device itself may become submerged and unusable due to water damage. What is common in these disasters is
It is very likely that the entire storage device will be damaged and will fail at the same time, making it unusable. In other words, in such a disaster, even if the conventional method is simply used for mirroring or duplexing, the effect can hardly be expected. The performance required for the original backup method must be such that at least one piece of data can be secured regardless of the situation of the computer system (hardware) in which the data was used. Therefore, the backup method according to the present invention has been devised in order to reliably back up data even in the event of a disaster such as that described above.

[0003]

SUMMARY OF THE INVENTION An object of the present invention is to provide a method for solving the above problems and hardware effective for backup measures.

[0004]

In order to achieve the above object, in the backup method in the computer network system of the present invention, when the data write / save operation is performed in the system, the same data is always stored in a plurality of storage devices. The feature is that distributed storage processing is performed, and when this is performed, not only data is distributed, but at the same time, relational data distribution is performed to a safer place. In hardware, we devised a control method that allows multiple SCSI storage devices to perform the same operation at the same time by a method different from the conventional mirroring device,
According to this method, a very inexpensive mirroring device can be supplied.

[0005]

[Operation] By performing the distributed storage operation of the same data at the same time and toward a plurality of safer remote locations, the probability that these distributed stored data will disappear at the same time becomes as close to zero as possible. A reliable backup of data becomes possible. Further, by supplying an inexpensive mirroring device, it becomes possible to use the mirroring device for the data at the end of the network system as well, thereby improving the backup performance in normal times.

[0006]

[Embodiment] First of all, the system with data link
If you explain about things, networking will be done
By the way, within the range of mutual input / output transfer of data.
If so, the route for the transfer is wired (telephone line, optical
Data link can be performed regardless of wireless (cable etc.)
It can be said to be a broken system. In the present invention, such a data
The data used in the linked system.
When performing data storage processing on the storage device, be sure to store two
Writes the same data to more storage devices
However, for example, this can be done with a minimum of two storage devices.
If you do, these two are the system that data link is done
Where on the link system is located in
Basically, it doesn't matter at all and can be arbitrary.
This will be described specifically with reference to FIG.
One of the storage devices where data is written is HD
At the time of D2, the other one can be HDD1 and H
It can be DD6, or when one is HDD3, other
One of them can be HDD5.
is there. However, the other one in this case is always local
The fact that it is a remote storage device outside the loop,
The write operations performed by the two
The important point is that it is done in a time that can be regarded as
It In addition, this behavior allows data from outside the local loop.
When writing data to a local storage device,
To the remote storage device outside the other local loop seen from
It is also possible to write one data. This
Outside the local loop, in other words on the remote loop
Including a storage device that stores two physical data
It means that the places are distributed to different places and stored in memory.
Performing at the same time means that two storage devices simultaneously
It is a means to minimize the probability of loss. But
However, in practice, no matter what method and means
It is impossible to do it in time processing (real time).
This is the multiple notation used in the data link system.
How the storage devices are connected on the system
The connection path and the basic performance of the data link itself,
For example, it depends on the data transfer rate. data
Is always done via some transmission path
It is necessary, and there is always a time delay in this transmission path.
Occurs. However, normal data link systems
The upper link is faster and larger than the lower link
Generally composed of a capacity link system
Therefore, even if this transmission path delay is included, two storage devices
To simultaneously start the processing to write to
If you say, it can be considered that it was done in near real time
Will be In the above, it is the most basic with two storage devices.
I explained about this idea, but as an actual business use
More storage in the system being used
Is used on the data link system, this
Flow of data in a system like
A more specific description of the processing operation performed by the terminal shown in FIG.
By. Figure 2 shows that the number of users is increasing.
Client-An example to explain the server system
System A, system B, system C
The server uses RAID-A, B, C as respective storage devices.
And the transmission line L1 between each server,
It is possible to transfer data mutually by L2 and L3.
It Each server is connected to multiple terminals that are clients.
The server S1 in the system A is connected to H
A-1, A2, A3 and HD that use DD-1 as a shared disk
A-4, A5 and HDD-3 with D-2 as a shared disk
A6, A7, A8, A9 which are shared disks
HDD4 is a shared disk for server S2 in system B
B1, B2 and HDD5 are shared disks
3, B4, B5 and HDD6 as a shared disk B6
B7 and B8 are H for the server S3 in system C
Both C1 and C2 that use DD7 as a shared disk and HDD8
Shares C3, C4, C5 and HDD9, which are used disks
Disks C6, C7 and HDD 10 are dedicated disks
C8 is connected. This network system
In conventional, such client-server system
System, server 2, server 2, server
Each server in server 3 is multiple servers in other locations
Via a transmission line for exchanging data with
Data distribution at the same time, so that
When the data of No. 1 becomes unusable, the data is transmitted via the transmission line L3.
Data from the server 2 or via the transmission line L1.
Be sure to use the data of server 3
One piece of data was to be secured. And
Server 1 is Hokkaido, server 2 is Tokyo, server 3 is 9
Earthquakes can occur when you are in a remote location, such as a state
In the event of another disaster, all data will be erased at the same time.
It seems unlikely that you will lose it. However, this
The data to be secured is the data already transferred to the server side.
It is only In other words, it is located in the bottom layer in FIG.
In each terminal such as personal computer or workstation
The data being created by various running applications.
In addition, as described above, up to the terminal of the lowest layer
It is rare to use RAID or disk duplexing
And the data created in this bottom layer is duplicated.
Means that the terminal user can
Data is transferred. For this reason, these various applications
When using application software, the user
Write the created data to HDD or FDD several times
Is normal, but usually when you do this
It's rare to write two. Therefore, at this time
If that data becomes unusable, it will be created by then.
The data created will be completely lost. Like a disaster
Data loss accidents that occur during normal times excluding abnormal situations
Most of this is due to the HDD crash during this work
Many are due to accidents, and the probability of occurrence is relatively high
Since it is a thing, take measures to respond to such accidents
Need to be kept. Therefore, in the method according to the present invention,
Layer terminal, for example, the data created in A1 in FIG.
If there is a data, and this data is written to HDD1,
In the conventional system, this data is stored in HDD1.
Although it was only written, according to the present invention
This method is the same as writing this data to HDD1.
Sometimes, this data is transferred to S1 and this data is transferred to S1.
Data to RAID-A and the data to be written is duplicated. It
In addition, S1 is S2 or S3 via the transmission lines L1 and L3, or
Transfer this data to both S2 and S3 and receive it
S2 is RAID-B and S3 is RAID-C.
By writing data, S1, S2, S
The same duplicated data is secured for each RAID
Be done. Another example of data distribution is system B.
The data created in the lowest terminal B1
When writing to HDD4, write this data to HDD4
At the same time, B transferred to S2 and B5 and received it
5 writes this data to HDD 5, S2 is RAID
At the same time as writing this data in -B, the transmission lines L3, L
Transfer data to S1 and S3 via 2 and receive it
S1 writes this data to RAID-A, and S3
Writes this data to RAID-C. In some cases
For example, S3 writes the data to RAID-C.
At the same time, transfer this data to C1 and receive the data.
The retrieved C1 is said to write this data to the HDD7.
Things can be done. Each of these terminals
To a local disk, including shared disks that are connected
The operation when writing data is the same as writing.
Whether to transfer data or not
Is whether or not to write the same data.
If you want to do this, first specify where you want to transfer the data.
By setting it to, the software that determines the operation of each terminal
Software for each server and client in this example.
It should be installed in the terminal that will be the ant, currently used
If you have a computer or workstation that has
Since the task task operation is a common method, the present invention
Even when incorporating the method of
Rather than installing this software for each application software,
Software that runs at the same time as application software
It is considered desirable to incorporate it as. In addition, each end
Transfers data depending on whether the end has performed a write operation
Whether or not the device device of the OS used at that time
Can be detected only by software by examining the liper area
However, in this case, it is necessary to pay attention to the type of OS
Because the address of the device driver area differs depending on
Yes, all OS that may be used in the system
Of course, it is necessary to deal with next,
Example of using this method when actual software is used
Will be described. In the above explanation, this method is
When using, the operation of each terminal is
It is said that it is performed according to the conditions set in the wear
The software for setting these conditions was used.
A screen display example of A is shown in FIG. First of all on the server side
To explain the software used, each setting is a keyboard.
Keyboard input or use the mouse controller
Cursor to the setting item position and click
It is assumed that the system used in Fig. 2 is used as an example.
If you clarify, this server will first enter the server name,
Software is used on Server 1 of System A
If it is one, enter S1 here. Next,
Enter the number of local discs. This is S1 low
The number of drives that will be KALDISK becomes 1, and then
Enter the drive name of RAID-A. If S1
If there are multiple local disks, RAID-1, R
AID-2, ... All drives like RAID-N
Enter the Eve name. Next is the main drive name,
This is not available if you have multiple local drives.
When writing data from outside Karloop to which drive
This is a setting for deciding whether to write, and is an example in Figure 2.
Since there is only one local drive, RAID-
It becomes 1. This next setting is connected to server S1
The number of groups with clients and their main dry
Settings of the client's group in this example.
The number of loops is 3. And the main of each group
The drive name is A1, A2, A3, H in place of group A =
DD1. A4, A5, HDD2 in place of group B =. Gu
A6, A7, A8, A9, HDD3 in place of loop C =
input. Do the same for other servers,
Software used in S1, S2, S3 on this server side
When the software settings have been completed up to this point,
In some servers, the setting data between servers
Data exchange and create a map of the network system.
The software for creating the □ map of Figure 3
By moving the mouse cursor to and clicking
To start. This allows each server to
By having a map of the stem, where and what dry
It can be used to know if there is a bump. This map
The following settings to be made after creation are the local drive of the server.
When data is written to the
Set whether to write the same data to the disk
If this is NO, the server will
Data will not be retransmitted, and when the answer is YES
Used on the network system as shown in Fig. 4 on the screen
All the disk names (storage device names) that have been recorded are displayed.
It At this position of the displayed disc name, mouse cursor
By moving the cursor and clicking
Setting to transfer data to the
Set. This setting can be performed on multiple units at the same time,
If you do this, the same data will be transferred to multiple locations simultaneously and written.
Can be imprinted. In this setting
For example, the software used in S1
Is RAID-B and RAID-C
Data is transferred from S2 to S1 and this data is RAID
-When writing to A, data is transferred to S3 and RAID-
Let C write data, but what about S2?
Also, the processing of is not performed. That is, the settings are done
If you do not re-transfer the data transferred from the
That's what it means. This is the transfer over and over again
It is a process to prevent entering a permanent loop that ends up being absolutely
Is what you need. Used by S2 and S3 on other servers
If you configure the software for
All the settings for the software used on the side are complete. Figure
Setting software for other servers S2 and S3 used in
An example of setting software is shown in Fig. 5, Fig. 6, Fig. 7 and Fig. 8.
You Next, set the software to be used on the client side.
Will be explained. Use this on the client side
Set the software to be used on the server side software.
A) After setting and completion of installation. Well
Start the software for this setting on the client side
Then, the map data on the server side is automatically acquired.
To include. By importing this map data
The setting software on the client side is also the network system.
Be able to understand all drive names used on the system
become. After this loading is completed, the setting screen shown in FIG.
Surface display is performed. Of this client software
This is an example screen for setting, but in this case as well,
For setting, use keyboard input or mouse controller.
The mouse cursor to the setting item position and click
It is supposed to perform the operation. First, the setting is
Other remote drives when writing to a local disc.
YES or NO to write on live
select. If you want to see the system map at this time □
By clicking the map display, the map table as shown in Figure 2
You can also see the display screen. When this transfer setting is set to NO
Does not transfer or write data, and is set to YES
Displays all drive letters on the network system
And move the mouse cursor to the drive name position
Click to perform data transfer and writing.
You can specify the drive to be included. In this case also
You can configure multiple drives as well as server settings.
In addition, in this client side setting software
Data transferred from where the settings are made
Must be retransmitted.
Yes. Next, in the explanation of the above-mentioned conventional backup method,
Disc mirroring featured in and how to do this
About computer mirroring device
The reason why it is often used on the main computer
I will explain again about the disk mirroring method
For normal use of the computer system
In addition, it is a very effective data backup method.
All storage devices used in the system if possible
It is desirable to use this in the storage. But
However, in reality, for example, a client-server system
In the case of network OS software such as
Disk mirroring and disk duplexing features
Even if the software includes, the function is effective
If the server side storage device only, or software
Software that does not include a mirroring function
In many cases, it is limited by the
In hardware, this mirroring device itself
This is a very expensive thing, so
It is rare to use it as a storage device at the end of a computer system.
It's impossible. Therefore, it was devised according to the present invention
Write the same data to multiple SCSI storage devices at the same time
Controls that allow loading or reading
Method and SCSI device controller using it
It By using the method according to the invention, a mirroring device
Can be supplied at a very low price. Below, this one
At present, a computer device has an HDD or an external storage device.
Best when connecting an optical disk drive
The interface used is SCSI. This
When inputting / outputting data in the SCSI system of
There are two types of transfer method, asynchronous transfer method and current transfer method.
Almost all the SCSI devices that are available can transfer data at high speed.
It is compatible with an effective synchronous transfer method. Synchronous transfer method
Initiates when the computer is powered on or reset
Synchronization messages between the target and target.
By confirming the transfer rate and offset value,
The settings are made so that they can correspond to each other. This, S
Size of SCSI device used in CSI method
The unique feature is that it is an intelligent device
Can be given. This is a traditional floppy disk
The disk that was run by the CPU on the computer side
There is no need to control the drive head position, etc.
All you have to do is transfer the mand. by this
The load on the CPU of the computer is reduced
Therefore, it is very efficient when the HDD is used in normal usage
Is a good way to do this, but a SCSI device to do this
Each has a built-in CPU and is executing commands
Will operate at a timing independent of each other. This
Control initiative when reading and writing data by this
However, in the case of a floppy disk,
However, when using a SCSI HDD,
In some cases, the HDD takes the initiative in reading and writing operations, and HD
D is the timing from the command from the computer side
To control the SCSI bus. This means
Two or more SCSI devices at the same time
If you try to do the same operation, there is a big problem
Become. For example, send the same command to two HDDs at the same time.
However, each HDD has a separate CPU with a separate CPU.
Because the processing by imming is performed, the timing
If you include it, the operation will start at the same timing.
It is impossible to finish. 2 because of this
HDDs are set to the same ID number and connected to the SCSI bus.
Even if they continue, they cannot be operated at the same time. SCSI
In the method, normal commands and messages are general R
Asynchronous transfer by handshake with EQ and ACK signals
Data transfer and synchronous data transfer can be used
Data transfer phase only. Therefore, in the present invention
Multiple SCSI devices to perform the same operation at the same time
The method is to use multiple SCSI devices during asynchronous operation.
A method to match the operation timing of the chair and when synchronizing
The operation timing of multiple SCSI devices.
How to make. And are used in combination. This way
If you explain about the actual example,
HDD as a SCSI storage device to save
The number of units is the same and the number of units is the same.
Explain about how to write and read the work
Do the light. First of all, in asynchronous operation, two H
A method for causing the DDs to simultaneously perform the same operation will be described.
As described above, two HDDs have the same ID number.
Even if the same command is given at the same time
HDD sends a signal for the response to this command.
The output timing is completely random and each HDD
Are A and B, A starts the operation first and outputs a response signal.
In some cases, B starts the operation first and sends a response signal.
Sometimes output. Bus is used in SCSI method
What can be done is in the device connected to the SCSI bus
, Only one of them can be used by multiple devices simultaneously.
Since it cannot be used for HD, the HD that started operation first
The response signal of D is output to the SCSI bus as it is.
And the device took the priority of the bus,
One HDD is left as it is.
U. Therefore, HDD output signals REQ, BUSY, MS
The G, C / D, and I / O signals are not shown in FIG.
This is solved by the timing control circuit. Figure 10
-1 is a circuit example when two devices are used,
FIG. 10-2 shows an example of input / output signals of this circuit. Moreover, this
Even if more devices are used, the circuit of Fig. 11
It is possible by the method shown in the example. The operation of FIG.
As can be seen from Fig. 10-2, the inputs A and B of this circuit
When the input signals of both become the same "H" or "L"
Only the output signal C corresponds to the "H" or "L" of the input signal.
Change accordingly. That is, A or B when the output is "H"
Even if one of them becomes "L", it is not output and both A and B are "
The output signal C becomes "L" after it becomes "L".
From the state where the output C of "L" becomes "A" or "B"
The output does not change even when becomes "H", and both A and B become "
The output C will return to "H" when it becomes "H".
5 circuits are created, and inputs A and B are HDD-A and
Output signals REQ, BUSY, MSG, C / of HDD-B
Input D, I / O, and output signal C
If connected to the SCSI bus on the side of the sita
Both A and B inputs to "L" or "H" at the same time
Drive the SCSI bus only after that
After two HDDs A and B are in the same operating state,
Status of other SCSI devices, mainly the initiator
Will be output. For example, for some command
HDD-A outputs a REQ signal, and HDD-
Initiator side SCSI when B outputs REQ signal
REQ of the bus HDD-B outputs the REQ signal
Is not driven until the
Must always return two HDDs as a response signal.
Is the timing after the REQ signal is output.
Therefore, it is not possible that only one of the HDDs works.
There is no pair. Other, BUSY, C / D, I / O, MSG
Similarly, the operation of the two HDDs must be the same.
This is because we will not drive the SCSI bus until later.
However, it is the same when receiving the response signal from the initiator.
Will be performed in the operating state. Control signal line
Among them, signals ATN and AC output from the initiator side
K and RST are each H directly or through a buffer.
You only need to connect it to the DD, data bus and parity bar.
The I / O signal of the SCSI bus on the initiator side
The direction of the input / output is detected, and H is detected from the initiator side.
When data is input to the DD, this data is stored in both HDDs.
Data should be input and output from the HDD side
If you want to use only one data, the initiator side
It should be set as the one to be output to the SCSI bus. This
However, the two HDDs used in this case have the same ID number.
In addition to setting the same,
The area reservation capacity and physical sector length are the same, and the data
The contents must be the same. Works asynchronously
For SCSI devices that do not
Although it is possible to perform the same operation at the same time,
Akinotori Most of the SCSI devices currently in use
Is compatible with the synchronous data transfer method.
Correspondence to the synchronous data transfer method becomes indispensable. Since
The method corresponding to this will be described below. Sync Day
In the data transfer method, the read data from the HDD is initialized.
When transferring to the initiator and data from the initiator side
May be written to the HDD. In the present invention, 2
Write or read data from one HDD at a time
However, when performing a read operation,
While both HDDs are operating normally
Is the same because both HDDs output the same data.
Make only one unit read and perform the operation and read the data.
It is also possible to transfer to the initiator, but HDD
In the case of
I don't know unless I read the data. others
Therefore, in the present invention, data is read from the HDD.
In the case of the above, the same operation is performed on the two HDDs at the same time.
And then initiate only one of the data.
It is supposed to be transferred to the data. At this time, the two HDD
Which data to use depends on the data read
Data of the HDD whose response signal output is delayed
To use. The reason for this is the movement of the head between the two HDDs and
It is said that the same operation is performed at the same time including the position etc.
Sometimes this method uses which HDD data
Also by a random selection by deciding what to do
The data actually read next to is transferred to the initiator.
By sending and using it, the data on the two HDDs will be correct.
Check whether it is saved in a state where it can be read out at all times.
There is also an advantage that the check can be performed directly. Sync data
When reading data by the transfer method,
The two different methods are used in a different way from the asynchronous operation described above.
It is necessary to match the operation timing of the HDD. Below
I will explain this. First, in the synchronous data transfer method,
When reading data, the REQ signal and A
The relationship between the CK signals is briefly explained as follows.
HD that received the data read command from the data side
D REQ data within the range that does not exceed the offset value
Prior to the ACK pulse at the timing synchronized with the pulse
Can be transferred to the initiator and received normally.
When taken, the initiator sends a request signal for the next data.
And outputs an ACK pulse to the HDD.
The number of ACK pulses is the REQ pulse output from the HDD.
It will never be output in excess of the number of scans.
If ACK pulses more than the pulse are returned, H
DD stores this as an error.
It will be pushed. From this fact, in the present invention, this R
It is necessary to control the EQ pulse and the ACK pulse.
Initiator in HDD that supports synchronous data transfer
After the synchronous message exchange between
The two HDDs (A, B) are simultaneously de-selected from the initiator.
Asynchronous until receiving data read command
Change from operation to synchronous operation. First, the two HDDs are
The offset agreed in the synchronous message exchange.
The number of data less than the threshold value is initiated with the REQ pulse.
The output data to the
There is a time lag. At this time, of the two HDDs
SCS is the REQ pulse and data output before
The REQ pulse of the person who did not output to the I bus but did output later
Data and data to be output to the SCSI bus.
In order to do this, which of the two HDDs will output first
REQ PAL that was output with a delay after detecting whether the work was performed
The output of the REQ pulse detection is the REQ pulse detection.
It is an output circuit. This is obtained in the REQ pulse detection circuit
Signals SEL-A and SEL-B are used HD
Control signals BUSY, MS other than D data and REQ
Used to select G, C / D, I / O and output to the bus
This is the select signal used. This REQ pulse detection
An example of the circuit is shown in FIG. In addition, other control signals
R output by another HDD whose data is not used
The processing and processing of the signal output from the EQ pulse and the initiator
And the control method will be explained above.
As shown in the HDD, the ACK pulse of more than the output REQ pulse
Error is entered when a password is entered.
Two H's trying to perform the same operation at the same time.
Control method that prevents the DD from entering an error state
Devised a law. This will be explained with reference to FIGS. 13 and 14.
U. FIG. 13 shows data reading in the synchronous data transfer method.
REQ pulse and ACK pulse
3 is a block diagram of a circuit for processing and controlling
It This circuit is composed of 1.2 HDDs, which is the unselected HDD
For counting the REQ pulse output by the
-Circuit 1 2. To count ACK pulses output to HDD
Counter circuit 2 3. Count the ACK pulses output by the initiator
Counter circuit for saving 3 4. A for generating ACK pulse to output to HDD
CK pulse generator circuit and its reference clock circuit and A
A mass for correcting the waveform of the output of the CK pulse generation circuit
Circuit 5. Compare the count values of the counter circuits 1 and 2 and
5. Comparator circuit 1 for detecting a match Compare the count values of the counter circuits 2 and 3 and
Comparator circuit 2 for detection of match Detects the level of the ACK signal output by the initiator
However, it is composed of a level detection circuit for controlling the mask circuit. For the two HDDs described above
Select and output any one of the data and REQ pulse
The HDD of the non-selected side
Input the REQ pulse to the counter circuit 1.
To count. The counter circuit 2 will be explained later.
Input the output of the ACK pulse generation circuit that performs
To install. The output of the initiator is output to the counter circuit 3.
ACK pulse is input and this is counted. Each coun
A length of about 8 BIT is usually sufficient. this
Indicates that the difference in the count value between each counter is less than the offset value.
There is no exception except when an error occurs, and the count is over.
-When the counter value of each counter became the same
Because it can detect the match without any problem by detecting
is there. Comparator circuit 1 is counter circuit 1 and cow
The match value of the count value of the input circuit 2 is detected, and the comparator
The counter circuit 2 includes the counter circuit 2 and the counter circuit 3.
This comparator circuit 1 detects the coincidence of count values.
Match detection was performed in either or both of
In this case, the operation of the ACK pulse generating circuit is stopped. This
Controlled by the counter circuit and comparator circuit of
The number of output pulses of the generated ACK pulse generation circuit is 2
Do not exceed the number of REQ pulses of either HDD.
It However, the ACK pulse created by the above method
Be used as it is as an ACK pulse for two HDDs.
I can't. The reason for this is that the initiator
The ACK pulse to be output is a data block delimiter,
Notify the HDD of the end and prepare for the next operation.
In order not only in the number of pulses but in the level as well.
The output is different from usual. Specifically, the ACK pulse
Normally, after leaving the last data, leave it at "L".
It is to return to "H" level. Therefore A
Even when creating a CK pulse and inputting it to the HDD
If you use ACK pulses that are just a combination of numbers, you get 2.
The HDD does not move to the next operation at different timing.
The operating status is different, resulting in an error status.
Operation stops. Books to deal with this
The invention uses the following method. Initiator side output
The state that the continuous "L" is not included in the applied ACK pulse
Then, the two HDDs are controlled by the above-mentioned controlled ACK pulse.
The same operation is performed with the number of read data controlled.
Therefore, the two HDDs output the last data.
Timing is also the same, and ACK pulse is generated at this time
The ACK pulse timing created by the circuit is also initiated.
It is determined by the ACK pulse output by the data. At this time,
The ACK pulse output by the initiator by the bell detection circuit
Whether or not the "L" level has been output for a certain period of time or more
Is detected, and when this is detected, this detection signal
Activate the mask circuit. The mask circuit is an ACK pulse
Masking the ACK pulse created by the pulse generation circuit
Of the normal ACK pulse output by the initiator
This mask circuit converts the waveform according to the format.
By inputting the output to two HDDs as an ACK signal
2 HDDs will move to the next operation at the same time.
It The operation timing of this mask circuit is shown in FIG.
However, in FIG. 15, for detecting "L" for a certain time or longer
Level detection circuit, mask circuit, ACK pulse generation circuit
For example: According to the circuit and control method described above
By inputting the created ACK pulse to two HDDs
However, this method is the same.
It is a control method at the time of reading at the time of data transfer
At the same time when the read command is finished, it is input to the HDD.
The ACK pulse from the initiator to the ACK signal from the initiator
Also, the output signal from the HDD is not the selected signal.
The output of the timing processing circuit during asynchronous
Switch so that it is output to the bus on the
It is necessary to return to the production state. Next, during synchronous data transfer
Write the same data to 2 HDDs at the same time
The method of performing the operation will be described. First, sync data
To explain the write operation during data transfer, HD
When D receives a write command from the initiator,
The number of REQ pallets in the range that the HDD does not exceed the offset value
Output. Initiator for this REQ pulse
The same number of data and the output timing of this data
The synchronized ACK pulse is output to the HDD. 2 at this time
HDDs output REQ pulse at the same timing
In this case, you can simply connect in parallel, but this
In the case of read operation as well,
The responses have random lags, so they are just
The HDD whose operation is delayed even if the columns are connected is the REQ pulse
The ACK signal is input before the
Number of ACK pulses ≤ REQ
It is out of the ruth number and an error occurs.
It is. In the present invention, as a method for solving this,
ACK input to the HDD at the time of reading
I created a pulse and controlled two HDDs.
In case of only, the HDD can write to the initiator
Available data, or a signal requesting its transfer.
Controls and outputs the REQ pulse that becomes the signal. 2 units
From the REQ pulse output from the HDD of the
The number of REQ packets that can be written by both D
Create a rule and send it to the REQ bus on the initiator side.
Input, the initiator outputs data.
It controls the ming. The circuit to do this
The block diagram is shown in FIG. The operation of this circuit is shown in FIG.
According to the explanation, it can be seen from FIG.
The circuit configuration is the same as the one described above for reading data.
It is very similar to the control circuit. The difference is
Mask circuit and level detection to control it
There is no circuit, and the signals input to each counter circuit are different.
The ACK pulse generator is the REQ pulse generator.
Therefore, the control of this REQ pulse generation circuit is
As before, the line is detected by the match detection output of the comparator circuit.
The circuit or its operation is basically the same.
Therefore, it is said that it is one thing for the sake of explanation
Generally, the circuit is shared and used.
In this writing circuit, the counter circuit 1 and the counter
The input circuit 3 inputs the REQ signal output of each of the two HDDs.
Each is counted as a force signal. Counter times
Path 2 receives the output signal of the REQ pulse generation circuit and
Count the output pulse of. Of the counter circuit 1
Comparing the count value and the count value of the counter circuit 2
The data is compared by the starter circuit A to detect a match. Also counter
The count value of circuit 2 and the count value of counter circuit 3
Matches are detected by comparing with a comparator circuit. This
Match detection in either or both of the comparator circuits A and B
The operation of the REQ pulse generation circuit is stopped while
It shall be. Output by this circuit and control
The number of REQ pulses is the number of REQ pulses output by two HDDs.
Output the same number of REQ pulses as the lesser of the
It will be one. As a result, this REO pulse generation circuit
Output to the initiator's SCSI bus.
Both HDDs can write to the initiator.
2 units because only a large number of data and ACK pulse are issued
HDDs of the same
You can write data. Place for writing data
In the case of two, the data output from the initiator side is 2 units.
Is input to both HDDs, and from other initiators
Is also input to both HDDs. Output from HDD
As for the force signal, only the REQ signal is as described in the counter circuit.
The other signal output is input to the data read
The initiator output signal of the selected HDD.
Data was written to the bus
Is the output signal of the timing processing circuit when asynchronous.
It shall be output to the SCSI bus on the shear side. This
However, this method is also a control method during writing during synchronous data transfer.
This is the same as when the HDD finishes the data write command.
Sometimes the REQ pulse input to the initiator is also asynchronous
When switching to the one of the timing processing circuit,
The issue must also be in its original operating state.
For example, in the circuit example and the control method described above,
For example, if the control signals BUSY and SEL of the SCSI bus are BU
For SY, as an output signal of HDD and for SEL
Is explained as an output signal on the initiator side.
But. This is to clarify the explanation of the operation, and the book
Coming Busy and SEL are bidirectional buses. So this
When BUSY and SEL are used as bidirectional input / output circuits
A circuit example is shown in FIG. The above method, times
The two HDDs used by the road are from the initiator side.
This user can be regarded as one HDD,
The software can be used as is and operated
Since there is no change in the
It can be used without being aware of nothing, and it can be used as a conventional HDD
Backup copy of written data to FDD every time
You'll be freed from the hassle of having to save it.
It In addition, in the above explanation,
Since it is not explained, the following explanation
Connect two HDDs to the circuit described above, and
To both HDDs by giving a command from the data
The same data is written at the same time. Also, are the same at the same time
The operation is made to read the data, but this basic
Even if only the circuits are
Read operation is performed by random selection
= Data status check operation is in progress. And this
If the operation of both HDDs is not the same in the method, H
This is the case where some abnormality occurs in the DD. Movement at this time
The operation is the next operation regardless of which HDD has an abnormality.
You can use it because it stops without going into a state.
The user can immediately know that an abnormality has occurred in the HDD.
Wear. Even at this time, either HDD is always operating normally
You should be able to do this, so the HD that works normally
D directly to the initiator side without passing through the circuit according to the present invention.
You can use the normal method of connecting to the SCSI bus of
It will be a bad thing. In this case, if an error occurs,
Only the basic circuit that does nothing to the subsequent processing
However, in reality it is inconvenient.
Therefore, for example, in the circuit according to the present invention, the operation of two HDDs is performed.
If there is a difference in the
Since it exists, this is BUSY, MSG, I of two HDDs.
EX-OR (NOR) for each signal such as / O, C / D
If you compare, there is a difference between the two signals for a certain period of time.
If an abnormality occurs, it can be detected by simply using a buzzer, lamp, etc.
In addition to notifying the user of the abnormality, whichever of the two HDDs
Can be switched with a switch and connected to the initiator side
If you make a circuit like this and switch this
Can be used, and data parity
-Which of the two HDDs is available if a check circuit is added?
Since it is possible to detect that the HDD is in an error state,
This signal automatically activates a normal HDD initiator
It is also possible to connect to

[0007]

EFFECTS OF THE INVENTION The SCSI device controller according to the present invention is not only for large-scale networked systems for business use, but also when used for data backup for personal use, it is free from bothersome backup work and provides reliable backup. This is a possible method. Since it can be supplied at a low price, it can be used in a wide range from personal to business systems, and for business purposes, duplication of data up to terminal terminals can be realized with a low budget. When used together with the backup method by
The reliability of data backup can be further improved.

[Brief description of drawings]

FIG. 1 is an example of a computer system for explaining the basic idea of the present invention.

FIG. 2 is an example of a network system for explaining a specific embodiment of the present invention.

FIG. 3 is an example of a display screen of server software when software is implemented to use the backup method according to the present invention.

FIG. 4 is an example of a display screen when setting a data transfer destination in the server software.

FIG. 5 is an example of a setting display screen of server software on the server of S2.

FIG. 6 is an example of a data transfer destination setting display screen of the server software on the server in S2.

FIG. 7 is an example of a setting display screen of server software in the server of S3.

FIG. 8 is an example of a data transfer destination setting display screen of the server software in the server of S3.

FIG. 9 is a display screen example of client setting software when the backup method according to the present invention is implemented as software.

FIG. 10 is a circuit diagram of an asynchronous timing control circuit. An example of input / output signals of the asynchronous timing control circuit.

FIG. 11 is a circuit example when a plurality of units are controlled by an asynchronous timing control circuit.

FIG. 12 is a circuit example of a REQ pulse detection circuit.

FIG. 13 is a block diagram for performing control for reading data in the synchronous data transfer method.

FIG. 14 is a mask circuit timing diagram.

FIG. 15 is a circuit example of a level detection circuit, a mask circuit, and a REQ pulse generation circuit.

FIG. 16 is a block diagram for performing control for data writing in the synchronous data transfer method.

FIG. 17 is a circuit example in the case where the input / output buses are bidirectional.

Claims (8)

[Claims] 1. A computer network system in which a data link capable of mutually transferring data is performed using a wired or wireless transmission path, and any one of a plurality of storage devices included in the computer network system. When a computer performs a write / save operation of data, a computer device that forms a local loop in relation to this storage device has at least one remote storage device that is provided with a write operation to the local storage device outside the local loop. , The same data is simultaneously written to two or more storage devices included in the remote storage device side, in which the write operation is performed via the data link. The computer device that forms the local loop , When performing a write operation to a storage device which is a local storage device from the viewpoint of this computer device, the write operation is performed simultaneously to two or more storage devices including at least one remote storage device provided outside the local loop, A data backup method characterized by performing relational data distribution in which the same data is written. 2. The storage device according to claim 1, wherein the storage device provided outside the local loop is any storage device usable in the computer network system, and its type is not limited. Data backup method. 3. The data backup method according to claim 1, wherein the same data writing operation is performed in real time except for a time delay which is unavoidably caused by the use of the computer network system. . 4. A SCSI device controller used by connecting between an initiator and a target,
SCS for connecting the SCSI bus on the initiator side
A plurality of S for connecting the I interface circuit unit and a plurality of target SCSI storage devices
The CSI interface circuit section and the control circuit section for controlling the input / output signals of these SCSI interface circuits are provided, and each of the SCSI interface circuits on the plural side does not have a means for synchronizing with each other. When SCSI storage devices are set to the same ID number and connected, a command from the initiator side controls the initiator and multiple targets to simultaneously perform the same operation for multiple SCSI storage devices with the same ID number. Multiple S depending on the car
A control method for a SCSI device controller, which enables simultaneous reading of data by simultaneously writing the same data to a CSI storage device and performing the same operation, and a SCSI device controller using the same. 5. The control for causing a plurality of SCSI storage devices having the same ID number to simultaneously perform the same operation according to claim 4, wherein a command on the initiator side is simultaneously given to the plurality of SCSI storage devices, Characterized in that a control signal having a common timing is created in a plurality of SCSI storage devices with reference to the operation timing of the SCSI storage device that takes the longest time to respond to all the operation responses, and the control is performed by this signal. Control method in SCSI device controller and SCSI device controller using the same. 6. The control signal having a common timing in a plurality of SCSI storage devices according to claim 5, during a write operation by the synchronous data transfer method, during a read operation by the synchronous data transfer method, and in other cases. Different control signals are used during operation. By creating control signals corresponding to each operation and switching between these signals, it is possible to support both synchronous and asynchronous SCSI devices. And a control method for a SCSI device controller, and a SCSI device controller using the same. 7. The means for creating a control signal corresponding to each operation according to claim 6, for counting request pulses output by each of a plurality N of SCSI devices in a write operation by a synchronous data transfer method. N counter circuits C _{1 to} C _N and a request pulse generation circuit for generating a request pulse to be output to the initiator, a counter circuit C _P for counting the output request pulse, and this counter circuit C _P Count value and the counter circuits C _{1 to} C _N
Of N comparator circuits X _{1 to} X _N for comparing each count value of each of the comparator circuits X _{1 to} X _N and detecting the coincidence in any one or more of the comparator circuits X _{1 to} X _N. While the operation is being performed, the operation of the request pulse generating circuit shall be stopped, and the output signal of this request pulse generating circuit shall be a request pulse which becomes a common timing for a plurality of SCSI storage devices, and this should be sent to the initiator side. The control method is characterized by causing the plurality of N SCSI storage devices to simultaneously write the same data when the synchronous data is transferred. 8. The means for creating a control signal corresponding to each operation according to claim 6, in response to a command from an initiator side of a plurality N of SCSI storage devices in a read operation by the synchronous data transfer method. Order detection circuit for detecting order and data and REQ of SCSI storage device selected based on this detection signal
A select circuit for outputting a signal to the initiator side, counter circuits B _{1 to} B _N-1 for counting request pulses output by each device other than the selected device, and an acknowledge pulse output by the initiator. _A counter circuit B _A for counting, a level detection circuit for detecting the level of the acknowledge pulse output from the initiator in the time domain, a mask circuit controlled by this output signal, and output to a plurality of SCSI storage devices. Acknowledge pulse generation circuit for generating an acknowledge pulse, a counter circuit _BP for counting the number of output pulses, a count value of the counter circuit _BP , and the counter circuits B _{1 to} B _N-1 and B. each count value of the _a, compared Comprising a match N values detecting as many comparator circuits Y ₁ to Y _N, one of the comparator circuits Y ₁ to Y _N, or while the coincidence detection being performed in any more, the It is assumed that the operation of the acknowledge pulse generation circuit is stopped, the output pulse of this acknowledge pulse generation circuit is input to the mask circuit, the waveform is corrected, and this is used as an acknowledge pulse that becomes a common timing in a plurality of SCSI storage devices.
By giving to multiple SCSI storage devices at the same time,
A control method characterized in that a plurality of N SCSI storage devices are made to read data by the same operation at the same time during synchronous data transfer.