JPH08166909A - Storage controller - Google Patents

Abstract

(57) [Summary] [Purpose] Regarding backup of computer centers,
The purpose is to update the file of the sub center with a slight time delay with respect to the file update of the main center. [Structure] A file controller (hereinafter referred to as FCU) is provided with a line adapter or the like, and when a CPU requests update of a certain file, the update data is sent to the FC
Write to buffer memory (or disk) in U, and also FCU of sub center via the line adapter etc.
Send it directly to. The FCU on the sub-center side writes the updated data in its own buffer memory (or disk). Further, the main center side can read the copied file instead of the writing length of the file to the sub center. That is, the FCU-FCU data transfer is possible.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to backup of a computer center. A disaster such as an earthquake may occur in the area where the computer center is located, and the function of the center may be impaired. In order to prevent this, it is required that a sub-center be installed in an area different from the main center, and that the sub-center take over the function when the main center fails.

In this case, it is necessary to equip the sub center with a file equivalent to the file held by the main center.
The present invention relates to backing up such files.

[0003]

2. Description of the Related Art Equivalent files (hereinafter,
It is called "remote file". There are the following two methods for realizing (1).

.. File transfer by communication between CPUs (application programs) of both primary and secondary centers. ． Moving files by physically transportable media such as MT.

The creation of a remote file by the above-mentioned communication between CPUs will be briefly described below. FIG. 6 shows an example of conventional data transfer. In FIG. 6, 1a and 1b are main units, 2a and 2b are file control units (hereinafter referred to as "FCU"), 3a and 3b are disks, and 4a and 4b are communication control units (hereinafter referred to as "CCP"). 5a and 5b are terminals.

Each of the main units 1a and 1b has a C
Each of the FCUs 2a and 2b is composed of a PU 11a, 11b and a channel CH 12a, 12b, and each FCU 2a, 2b is a channel adapter (hereinafter referred to as "CA") 21a, 21b and a control unit 22.
a, 22b, device adapters (hereinafter referred to as "DA") 23a, 23b, buffer memory (hereinafter "BS")
Called. ) 24a, 24b.

The device configuration with the suffix "a" is the center side, and the device configuration with the suffix "b" is the sub center side. The remote file is created by the following procedure.

(1). The data input from the terminal 5a on the primary center side is written to the main storage device (not shown) of the main body device 1a via the CCP 4a. The CPU 11a processes the written data, writes the processing result to the main storage device, and then writes the processing result on the main storage device to the disk 3a via the FCU 2a via the CH 12a.
Ask.

(2). The control unit 22a of the FCU 2a is C
The write data is received via A21a, the write data is once written to the BS24a, and then the data is written to the disk 3a via the DA23a.

(3). From the viewpoint of the processing capacity of the CPU and the processing capacity of the CCP, the CPU 11a
That is, after the data is no longer input from the terminal 5a, the FCU 2a is requested to read the data to be backed up for one day or several days from the disk 3a and write it in the designated area on the main storage device via the CH 12a. To do.

(4). The control unit 22a of the FCU 2a reads out the designated data from the disk 3a via the DA 23a, writes it once in the BS 24a, and then writes it in the BS 2a.
The designated data on 4a is sent to CH12a via CA21a. The CH 12a writes the designated data in the area of the main memory designated by the CPU 11a.

(5). After the designated data is written in the main storage device, the CPU 11a sends the data to be backed up in the main storage device to the sub-center side C
Request CP4a.

(6). The CCP 4a that received the transmission request
The designated data is read from the main storage device via the CH 12a and transmitted to the CCP 4b on the sub center side via the transmission path.

(7). CCP4b on the sub-center side is CH1
It interrupts CPU11b via 2b. The interrupted CPU 11b designates an area of the main storage device (not shown) on the side of the sub-center for writing data to be backed up.

(8). The CH 12b receives the data to be backed up from the CCP 4b and writes it in the designated area. After the writing is completed, the CPU 11b designates the area and requests the FCU 2b to write the backup data to the disk 3b via the CH 12b.

(9). The control unit 22b of the FCU 2b is C
The write data is received via A21b, the write data is once written to the BS24b, and then the data is written to the disk 3b via the DA23b.

With the above, copying of the file is completed,
Remote file creation is complete.

[0018]

In the conventional method as described above, the contents of the file to be backed up in the main center and the contents of the file to be backed up in the main center are considered from the viewpoint of the processing capacity of the CPU and the processing capacity of the CCP.
There is a difference of one day to several days from the contents of the remote center remote file.

Therefore, when the main center encounters an accident, it is necessary to update the remote file of the sub center with the update data for one to several days and then start the operation of the sub center as the main center. , It took a long time.

Further, in the case of a write error to the disk on the side of the secondary center or a failure of the FCU, it is difficult for the application program on the side of the primary center to grasp which data has been backed up.

In view of the above point, the present invention has an object to provide a means for improving the reliability of a computer network by minimizing the time delay of the creation of the remote file of the sub-center.

[0022]

The above-mentioned problems can be solved by a storage control device configured as follows. FIG.
[FIG. 3] is a diagram illustrating the principle of the present invention.

1. First input / output unit 21 and control unit 22
A storage unit 24, a second input / output unit 23, and a third input / output unit 25, wherein the first input / output unit 21 is connected to another device. The second input / output unit 23 is an interface with another device, and the third input / output unit 25 is an interface with another device.
Is an interface with another device, and when a request for writing and sending data via the first input / output unit 21 is made in the storage control device 2, the control unit 22 To the storage unit 24, then read the data from the storage unit 24 and output the data via the second input / output unit 23, independently of or in parallel with the output operation. Output via the means 25, and a data write request is sent via the third input / output means 25 to the storage controller 2
If so, the control unit 22 writes the data in the storage unit 24, then reads the data from the storage unit 24 and outputs the data via the second input / output unit 23. When a data read request is issued to the storage control device 2 via the third input / output unit 25, the control unit 22 sends the data designated via the third input / output unit 25. Whether or not it exists in the storage unit 24 is checked, and if it exists, the designated data is read from the storage unit 24 and output via the third input / output unit 25. If it does not exist, the designated data is read through the second input / output unit 23, the read data is written in the storage unit 24, and then the designated data is stored in the storage unit 24. To read the third input / output unit 2
5 is configured to be output.

2. In the above 1, write order control means 2
7 is added and data is written and sent to the storage controller 2 via the first input / output unit 21, or data is sent via the third input / output unit 25. When the write request is sent to the storage control device 2, the write order control means 27 increases the priority of outputting the data from the second input / output means 23.

In the above-mentioned 1, data replacement control means 28 is added in place of the write order control means, and data write and transmission requests are sent to the storage control device 2 via the first input / output means 21. If there is, or if there is a data write request to the storage control device 2 via the third input / output unit 25, the control unit 22 writes the data in the storage unit 24. In this case, the data replacement control unit 28 may increase the priority of erasing the data from the storage unit 24.

Also. In the above 1, write order control means 2
7 and a data exchange control means 28 are added, and when a data writing and sending request is sent to the storage control device 2 via the first input / output means 21, or the third input / output means 21 is used. When a data write request is issued to the storage control device 2 via the output means 25, the control section 22 writes the data to the storage section 24. A configuration may be adopted in which the priority of eviction from the storage unit 24 is raised and the priority of outputting the data from the second input / output unit 23 by the write order control unit 27 is raised.

3. In the above 1, when the write completion reporting means 26 is added and a request for writing and sending the nth data is made to the storage control device 2 via the first input / output means 21, at least (n- 1) The condition that the second data is output from the second input / output unit 23 and a processing completion response to the output data is received via the second input / output unit 23, or the nth data is When any one of the completion conditions of the output condition from the third input / output unit 25 and the reception condition of the write completion report for the n-th data via the third input / output unit 25 is satisfied, The write completion reporting means 26 is configured to report the write completion of the n-th data via the first input / output means.

The following means can be adopted as the means related to the above means. First, in the above 1, the write completion reporting means 26 is added, and the third input / output means 25 is added.
When a data write request is made to the storage control device 2 via the, the write completion reporting means 26 reports the write completion of the data at the time when the writing of the data to the storage unit 24 is completed. Can be output via the third input / output means 25.

Next, in the above 1, when the write completion reporting means 26 is added and a data write request is made to the storage control device 2 via the third input / output means 25,
The data is output from the second input / output unit 23, and a processing completion response to the output data is output to the second input / output unit 23.
When it is received via the write completion report means 26
Sends the data write completion report to the third input / output unit 2
It is also possible to adopt a configuration in which the data is output via the signal 5.

Thirdly, in the above-mentioned 1, when the write completion reporting means 26 is added and a write request for the nth data is made to the storage control device 2 via the third input / output means 25. , At least the storage unit 2 for the n-th data
4 the completion condition of writing, or (n-1) th data is output from the second input / output unit 23, and a processing completion response to the output data is output via the second input / output unit 23. If either one of the completion conditions received by the above is satisfied, the write completion reporting means 26 sends a completion report as the write completion of the n-th data via the third input / output means 25. It is also possible to adopt a configuration in which the output is performed as a result.

In addition, in the above-mentioned 1, in the case where the write completion reporting means 26 is added and a request for writing and sending data via the first input / output means 21 is made in the storage control device 2, at least: The condition that the writing of the data into the storage unit 24 is completed, or the data is output from the third input / output unit 25, and then the write completion report of the data via the third input / output unit 25 is sent. The write completion reporting means 26 outputs a write completion of the data via the first input / output means 21 when either one of the reception conditions, that is, the completion condition is satisfied. It is possible.

[0032]

With the above configuration, that is, by providing the storage controller such as the file controller with the line adapter and the LAN adapter, the CPU and the CCP are not burdened with an extra load for updating the remote file. The remote file can be updated with a slight time difference from the file update.

Here, "without extra load"
Means that the CPU of the main center may request the subordinate file control device to update the data, and the CPU of the subcenter may be untouched for updating the subordinate remote file. In addition, "Data is not transferred via CCP in remote file update."
Means that.

Further, by providing the completion report means in the FCU, the normal end / abnormal end of the update of the remote file can be easily confirmed by the CPU on the main center side through the completion report means of the FCU on the sub-center side.

[0035]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention relating to remote file update will be described below with reference to FIGS. 2 is a block diagram of an embodiment of the present invention, FIG. 3 is a data eviction management (LRU) table of a buffer memory, FIG. 4 is a write order control (FIFO) table, and FIG. It is a completion signal generation circuit and a time chart.

Correspondences between the constituent elements of each invention in the principle explanatory diagram of the present invention of FIG. 1 and each part of FIGS. 2 to 5 are as follows. The storage control device 2 of FIG. 1 includes the FCUs 2c and 2d of FIG. 2 and the first input / output unit 21 of FIG.
1d The second input / output means 23 of FIG. 1 is the DA 23c, 2 of FIG.
3d The third input / output means 25 of FIG. 1 is the RA 25c, 2 of FIG.
5d, the control unit 22 of FIG. 1 has the control units 22c and 22d of FIG. 2, and the storage unit 24 of FIG. 1 has BSs 24c and 24d of FIG.
24c, d of the writing completion reporting means 26 of FIG. 1, the writing completion reporting means 26c, 26d of FIG. 2, and the writing order control means 27 of FIG. 5A, the writing order control means 27c, 27d of FIG. 4, the data exchange control means 28 of FIG. 1 is the data exchange control means 28c, 28d of FIG. 2 and the latest access time 9c, d of FIG.

In FIG. 2, reference numerals 1c and 1d denote main body devices and 2
c and 2d are FCU, 3c and 3d are disks, 4c and 4d
Is the CCP. The respective main units 1c and 1d have CPUs 11c and 11d and channels CH12c and 12d, respectively.
And each FCU 2c, 2d has a CA 21c, 21
d, control units 22c and 22d, DAs 23c and 23d, BS
24c, 24d, router adapter RA25c, 25d,
Completion reporting means 26c, 26d, write order control means 27
It is composed of c and 27d.

The device configuration with the suffix "c" is the center side, and the device configuration with the suffix "d" is the center side. FIG. 3 shows a detailed configuration of the BSs 24c and 24d, and both the primary and secondary centers have the same configuration.

BSs 24c and d display data 6c and d, read (hereinafter referred to as "R"), write (hereinafter referred to as "W") mode display 7c, d, and write completion in W mode. 8c, d and the latest access time 9c, d.

Each of the BSs 24c and 24d has a subdivided area for every k addresses and is read / written via each input / output unit in units of k addresses. In the write completion 8 of the BSs 24c and d, "0" means that the data has not been written to each of the disks 3c and d, and "1" means that the writing to the disk has been completed. In the R mode, this write completion 8 is meaningless.

At the access times 9c and d, the time ti at which each data is accessed is written every time the data is accessed. (The larger the value of i is, the newer the time is.) That is, this method uses LRU (Least Recentl).
y Used), which is a data eviction (data exchange) method of the memory.

For example, the k pieces of data 1 at the address “0 to k−1” of the BS are data accessed from an application program (hereinafter referred to as “app”) 1 and the latest access of the application 1 is made. The time is t1. Of course, even if there is an access from the application 1 before the time t1, it means that the latest access is at the time t1. Similarly, the k pieces of data 2 of the address “k to 2k−1” of the BS are data accessed from the application 2, and the latest access time of the application 2 is t3.

In this LRU system, the data with the latest access and the oldest is preferentially ejected (replaced). Therefore, in order to increase the priority of eviction (replacement), the contents of the access times 9c and 9d may be forcibly displayed as the old time. For example, in order to preferentially write the data 5 to the disks 3c and d, the content "t5" of the access time 9c and d of the data 5 may be changed to "t0".

Further, as a matter of course, the R / W mode 7
When c and d are in the W mode and data of which W completion 8c and d are "0" are expelled, new data is written to the area of the BS after the completion of writing the data to the disks 3c and d.

In the example of FIG. 3, data 1 is the oldest data. When the data 2 is to be expelled, new data is written to the addresses "k to 2k-1" after the writing of the data 2 to the disks 3c and d is completed.

The write order control (FIFO) table of FIG. 4 is a method of processing in a first-in first-out order, and stack address pointers 61c and d, stacks 62c and d, and the number of in-stack data 63c in each FCU 2c and d of both the primary and secondary centers. , D, and D in FCU 2c, d
A23c and d are the stack address pointer 61
BS2 according to the stack address values indicated by c and d
Data is read from the disks 4c and 4d and written on the disks 3c and d. When the writing is completed, the disks 3c and d send a writing completion signal to the DAs 23c and d.

When writing new data to the disks 3c and d, the control units 22c and 22d add the value of the stack address pointer 61 and the value of the number of in-stack data 63, and the value indicated by the addition result is added to the stack 62. The address of the BS 24 in which the new data is stored is written in the address of the stack 62 as the address of the.

After that, the control units 22c and 22 increment the value of the number of in-stack data 63c and d by "+1". Disk 3
To increase the write priority to c and d, stack 6
The addresses of the BSs 24c and d where the data is stored may be written in the address "0" of 2c and d, and the contents of the stack address pointers 61c and 61d may be set to "0".

In this case, the address "0" of the stacks 62c and d is normally left unused for emergency write data, and the initial values of the stack address pointers 61c and 61d are set to "1".

Alternatively, as another method, the stack 62
The address "0" of c and d is normally used, and when the emergency write occurs, the data of the address "0" of the stack 62c and d is shifted to the address "1", and the data of the address "1" is changed to the address "1". 2 ", and after sequentially shifting the number of data in the stack by the number indicated by 63c and d, the emergency write data BS24c,
There is a method of writing the storage address of d.

In the example of FIG. 4, k pieces of data from the BS address “4k” indicated by the stack address “0” are first written to the disk 3. FIG. 5 shows that when a remote file update request is sent from the main unit 1c of the main center to the FCU 2c, when the update of both the file on the main center side and the remote file is completed, the FCU 2c for which the update is requested is the main body of the request source. It is a circuit (a) for reporting the completion of updating of both files to the device 1c and time charts (b) to (e) of each signal. The same circuit as in FIG. 5 also exists in the FCU 2d on the sub center side.

In FIG. 5A, the logical function of the logic element 71c is the logical product "AND", and the logical function of the logic element 72c is the D type flip-flop "FF". The logic function of the logic element 71c may be "OR".

The inputs of the AND 71c are two inputs, one is a signal X, that is, a file update completion signal on the positive center side, and the other is a signal Y, that is, a remote file update completion signal.

The FF 72c operates at the falling edge of the clock signal, and its output is the update completion signal for the main body device 1c which is the update request source. Which signal X,
The logical value "1" of Y and Z also means that the update is completed.

In FIGS. 5B to 5E, the signal X becomes the logical value "1" (update completion) at the time T1 before the clock "1" is input to the FF 72, and the signal Y is the clock "3". Becomes a logical value "1" (update completion) at time T2 before being input to the FF 72. Therefore, the signal Z becomes the logical value "1" (update completion) at the falling time T3 of the clock "3". Based on the above operations and functions, the remote file is updated in the following procedure.

(1). The remote file update data input from the terminal (not shown) on the main center side is CCP4.
It is written in the main storage device (not shown) of the main body device 1a via c. The CPU 11c processes the written data, writes the processing result in the main storage device, and then requests the FCU 2c via the CH 12c to write the processing result in the main storage device into the disk 3c.

(2). The control unit 22c of the FCU 2c is C
The update data is received via A21c, and the update data is written to the address "4k-5k-1" of the BS 24c.
When writing the update data to the BS 24c, the control unit 22c performs any one of the following processes in order to raise the priority of writing to the disk 3c. (The current time of writing is t5.). This is a method using only the data replacement control means 28c as described above.

In this method, the write order control means 27c such as the stack 62c as shown in FIG. 4 does not exist, and the writing order is the oldest access time 9c. The control unit 22c forcibly sets the contents of the latest access time 9c to the oldest time t0, searches the access time 9c,
The write address is designated for the DA 23c in order from the oldest data.

Therefore, the control unit 22c specifies the write address "4k" to the DA 23c if the DA 23c is R / Wing the disk 3c at that time after the completion of the R / W. Become.

.. Writing order control means 27 as described above
This is a method using only c. In this method, the data exchange control means 28c does not exist, and the writing order is controlled only by the stack 62 shown in FIG. (Claim 2
(Correspondence) The control unit 22c forcibly writes the storage address "4k" of the update data to the address "0" of the stack 62c and sets the content of the stack address pointer 61c to "0". The content of the number of data in the stack 63 is changed from "1" to "2". In this method, some kind of data exchange means is required for R / W of general data other than remote file data.

.. The data exchange control means 28c
Although it has both means of writing order control means 27c, when writing the data sent from the main body apparatus 1c to the BS 24c, the latest access time 9c is set to t5 as shown in FIG. To do.

However, as shown in FIG. 4, the update data BS24c is forcibly placed at the address "0" of the stack 62.
The storage address "4k" is written to set the content of the stack address pointer 61c to "0". The content of the number of data in the stack 63c is changed from "1" to "2".

By this processing, the access time 9c is t5.
Therefore, the stack 6 that controls the writing order even though the unwritten data 2 (access time t2) exists
2c has the highest priority, so the update data is written to the disk 3c first. This method is effective when the update data for the remote file will be reused by the apps of both the primary and secondary centers in the near future. Because BS24
This is because if the update data exists in c and d, the update data can be sent to the CPUs 1c and d at high speed.

.. The data exchange control means 28c
And write order control means 27c, and both means are utilized. The latest access time 9c is forcibly set to the oldest time t0, and as shown in FIG. 4, the storage address "4k" of the update data is forcibly written to the address "0" of the stack 62, and the stack The content of the address pointer 61 is set to "0". The content of the number of data in the stack 63 is changed from "1" to "2".

This method is effective in the case where the update data for the remote file is not reused in the near future, contrary to the above. This is because even if the updated data exists in the BSs 24c and d, the control units 22c and d can preferentially use the area for new data thereafter.

BS24 can be obtained by any of the above methods.
After writing to c, the DA 23c writes the specified data to the disk 3c. When the writing is completed, the signal X is set to the logical value "1" as shown in FIG.

The method of the FCU on the main center side and the method of the FCU on the sub center side may be different methods. For example,
The method on the primary center side is the above, and the method on the secondary center side is the above method.

(3). As described in (2) above, the control unit 22c writes the update data to the BS 24c, DA
After instructing 23c to write the update data to the disk 3c, the address "4k" of the BS 24c is designated and R
The A25c is instructed to send the update data to the sub center side. Here, the temporal order relationship between the write command to the DA 23c of the control unit 22c and the send command to the RA 25c may be either first.

When writing to the disk 3c, a write error may occur and rewriting may be performed, and when data is transmitted from the RA 25c, a delivery error may occur and redelivery may be performed. , In general,
The writing operation to the disk 3c and the sending operation to the sub center side are executed independently.

Of course, the writing operation and the sending operation may be simultaneously performed in parallel for some data. (4). Upon receiving the remote file update data, the sub-center RA 25d notifies the control unit 22d that the data has been sent. The control unit 22d uses BS2
The contents of 4d are checked, the address of the BS 24d to which the received data is to be written is determined according to the LRU method, and the update data is written to the BS 24d in order from that address.

At this time, the control unit 22d checks whether or not it is the update of the remote file, and if it is the update of the remote file, the update data is sent to the BS 24d by any one of the methods (2) to (2). Write or
The contents of the stack 62d and the like are changed.

(5). The control unit 22d instructs the DA 23d to specify the address of the BS 24d and write it on the disk 3d. When the writing of the disk 3d is completed, the disk 3d notifies the control unit 22d of the completion of the writing via the DA 23d. The control unit 22d confirms that the writing of the remote file is completed, and then sends the writing completion signal to RA2.
Send from 5d to the main center side. On the positive center side, this write completion signal is the signal Y shown in FIG. As a means for generating the write completion signal Y, there is the following method other than the above.

[0073] Completion of writing to BS24d. When the nth update data is sent from the main center, the writing of the nth update data to the BS 24d and the writing of the (n-1) th update data to the disk 3d are completed (6). The RA 25c on the primary center side notifies the control unit 22c of the write completion signal Y on the secondary center side according to any one of the above methods (5).

As shown in FIG. 5, the controller 22c generates a write completion signal Z from the write completion signal X of its own device and the write completion signal Y of the other device, and requests the remote file update. The update completion is notified to the device 1c.

There are the following methods of generating the write completion signal X of the device itself. ． When there is a request to write and send the nth data via CHA, the (n-1) th data is sent to the disk 3c.
The signal X is set to "1" upon completion of writing in (corresponding to claim 3).

.. When there is a request to write and send the nth data, the signal X is set to "1" upon completion of writing the nth data to the BS 24c. Whatever it is,
If it is determined in advance which data is written and where the update is completed, the CPU (application program) 11c on the main center side can know up to what data the remote file has been updated.

In this embodiment, C on the sub center side
Although the update completion notification of the remote file is not sent to the PU (application program) 11d, it may be configured to notify. In this case, of course, the application program that handles the notification must be running.

[0078]

As is apparent from the above description, according to the present invention, regarding the backup of the computer center, the remote file of the sub center can be updated with a slight time delay with respect to the file update of the main center. Becomes

Further, the CPU (application program) on the main center side can know to what extent the remote file has been updated. in this way,
It has the industrial effect of increasing the reliability of protecting files on a computer network.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is a block diagram of an embodiment of the present invention.

[Fig. 3] Data eviction management of buffer memory (LR
U) Table

FIG. 4 Write Order Control (FIFO) table

FIG. 5: Write completion signal generation circuit and time chart

FIG. 6 Example of conventional data transfer

[Explanation of symbols]

1 1st other device 1a, 1b, 1c,
1d Main unit 2 Storage controller 20
Second other device 2a, 2b, 2c, 2d File control device 3 Second storage unit 3a, 3b, 3c,
3d disk 4a, 4b, 4c, 4d communication control unit 5a, 5b terminal 11a, 11b, 11c, 11d CPU 12a, 12b, 12c, 12d CH 21 first input / output unit 23 second input / output unit 21a, 21b, 21c, 21d CHA 22a, 22b, 22c, 22d, 22 control section 23a, 23b, 23c, 23d DA 24a, 24b, 24c, 24d BS 25 third input / output means 25c, 25
d RA 26c, 26d Write completion reporting means 27c, 27d Write order control means 28c, 28d Data exchange control means 6c, d Data 7c, d
R / W mode 8c, d W completion 9c, d Latest access time 61c, d Stack address pointer 62c, d stack 63 Number of data in stack 71c AND 72c F
F

Claims (3)

[Claims] 1. A storage control device comprising a first input / output unit, a control unit, a storage unit, a second input / output unit, and a third input / output unit. Input / output means is an interface with another device, the second input / output means is an interface with another device, and the third input / output means is an interface with another device. The interface is an interface, and when a request to write and send data is made to the storage control device via the first input / output means, the control unit writes the data to the storage unit, and then the storage unit. The data is read out and output via the second input / output unit and the third input / output unit, and a data write request is sent to the storage control device via the third input / output unit. If the control unit writes the data to the storage unit, the control unit then writes the data to the storage unit. If the storage control device receives a data read request via the third input / output means and outputs the data via the second input / output means, the control unit , The specified data is read from the storage unit, or is read via the second input / output unit and written in the storage unit, and then read from the storage unit, and is read via the third input / output unit. A storage control device characterized by outputting the output. 2. The method according to claim 1, wherein write order control means is added, and when a request for writing and sending data via the first input / output means is in the storage control device, or When a data write request is made to the storage control device via the third input / output means, the write order control means increases the priority of outputting the data from the second input / output means. Storage controller. 3. The method according to claim 1, further comprising write completion reporting means, and when a write and send request for nth data is made to the storage control device via the first input / output means, at least: The condition of outputting the (n-1) th data from the second input / output means and receiving a processing completion response to the output data via the second input / output means, or the nth data Is output from the third input / output unit, and thereafter, any one of the following completion conditions for receiving the write completion report for the n-th data via the third input / output unit is satisfied: The storage control device, wherein the write completion reporting means reports the write completion of the n-th data via the first input / output means.