JPH0651918A - Semiconductor disk device - Google Patents

Abstract

PURPOSE:To surely save data in a non-volatile medium, with respect to a semiconductor disk device. CONSTITUTION:Data stored in a semiconductor memory 3 allocated with plural logic drives are successively read for each logic drive and simultaneously written in plural disk devices 6-8 while being divided so as to save the data at this semiconductor disk device. This device is provided with a table 20 to record the number of times of data error generation with the fault generation of the disk devices 6-8 for each of disk devices 6-8 and to record a flag instructing disconnection for each of disk devices 6-8, and control means 21 to count time required for saving data for each logic drive when data save processing is continued by retry with the fault generation at the disk devices 6-8, and to disconnect the disk device, for which the number of times of data error generation is maximum, by referring to the table 20 when the total count time exceeds prescribed time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor disk device using a volatile semiconductor memory, and more particularly, it reliably executes data saving to a non-volatile medium for saving data provided corresponding to a power supply stoppage of the device. The present invention relates to a semiconductor disk device.

In recent years, as computer systems have become faster, there has been a demand for faster processing of external storage devices. For this reason, a semiconductor disk device has been provided which is capable of high speed access ten times or more as compared with a magnetic disk device.

However, since this semiconductor disk device uses a volatile semiconductor memory, the data stored in the semiconductor memory is erased when the power supply of the device is stopped. Therefore, the non-volatile medium is detected by detecting the stop of the power supply of the device.
For example, it is necessary for the magnetic disk device to reliably save the data stored in the semiconductor memory while the backup power supply is supplying power.

[0004]

2. Description of the Related Art FIG. 7 is a block diagram for explaining an example of a conventional technique, and FIG. 8 is a detailed block diagram of a disk controller.

In FIG. 7, the interface control unit 1 receives various commands from a channel (not shown) and transfers data to and from the channel, and also transfers data to and from the memory access control unit 2.

The memory access controller 2 includes a semiconductor memory 3
Is controlled to write or read data and transfer data to or from the interface control unit 1. Then, when there is an instruction from the control monitoring unit 4, data transfer with the disk control unit 5 is performed.

The disk control unit 5 controls the disk devices 6 to 8 based on the instruction from the control monitoring unit 4, divides the data sent by the memory access control unit 2 and writes the data into the disk devices 6 and 7, respectively. , The exclusive OR of each bit of the data written in the disk devices 6 and 7 is written in the disk device 8.

Then, in accordance with an instruction from the control monitoring section 4, the data read from the disk devices 6 and 7 is written into the semiconductor memory 3 via the memory access control section 2. At this time, if one of the disk device 6 and the disk device 7 has a failure, for example, if the disk device 6 is a failure, the disk control unit 5 determines from the data of the disk device 7 and the disk device 8 that The data stored in the disk device 6 is restored and written in the semiconductor memory 3.

The control monitoring section 4 controls the entire semiconductor disk device, such as detection of power failure, confirmation of data saving / restoring operation, and holding of the result. The backup power source 9 is composed of a battery or the like, and when the power source of the device (not shown) is cut off,
Interface controller 1, memory access controller 2, semiconductor memory 3, disk controller 5, disk devices 6-8
Also, a current is supplied to the control monitoring unit 4 for a predetermined time.

Normally, the interface controller 1 and the memory access controller 2 transfer data between the channel and the semiconductor memory 3, the data sent by the channel is written in the semiconductor memory 3, and the semiconductor memory 3 outputs the data. The read data is sent to the channel.

When the control / monitoring unit 4 detects the disconnection of the power source of the device, it instructs the disk control unit 5 to start the data saving operation. Therefore, the disk control unit 5 activates the data transfer sequence with the memory access control unit 2, and at the same time, the disk devices 6 to
The data transfer sequence with 8 is also activated.

Therefore, the data read from the semiconductor memory 3 by the memory access control unit 2 is written to the disk devices 6 and 7 via the disk control unit 5 as described above, and the data for data restoration is also used as the disk device. Written in 8.

Then, when the power is restored, the control monitoring unit 4
In accordance with this instruction, the data simultaneously read from the disk devices 6 and 7 are written in the semiconductor memory 3. The disk controller 5 has a configuration as shown in FIG. 8, and when a data save instruction from the control monitor 4 is issued to the disk controller 5,
The processor 17 detects this and activates the control circuit 18. The control circuit 18 sets the operation of the data managers 11 to 13 and the disk devices 6 to 8 and starts data transfer.

The data managers 11 and 12 are configured to automatically transfer data by hardware, divide the data sent from the memory access control unit 2 via the data bus 19 into buffer memories 14, respectively.
And 15 and write to the disk devices 6 and 7, respectively.

The data restoration circuit 10 receives the data from the data bus 19 and uses the data manager 1 as described above.
When an exclusive OR for each bit of the data divided by 1 and 12 and written in the disk devices 6 and 7 is created, the data is written in the disk device 8 via the data manager 13 and the buffer memory 16.

When the data is written in the semiconductor memory 3, if the disk device 6 is in trouble as described above, the data restoration circuit 10 sends the data read from the disk device 8 to the data manager 11, and the data manager 11 When the data to be written in the disk device 6 is restored from this data, the data is sent to the data bus 19.

The processor 17 includes the disk devices 6-8.
The data transfer status is monitored, and when the transfer is completed, a completion report is sent to the control monitoring unit 4. When a failure occurs in the disk device 6 or 7 at the time of writing data, the failure mode is a case where the subsequent processing cannot be continued due to a complete failure, the internal retry of the disk device 6 or 7, or the data manager 11 Alternatively, there is a case in which the process can be continued by the retry by 12.

[0018]

As described above, conventionally, there are cases where the failure mode of the disk device 6 or 7 cannot continue processing and cases where it is possible to continue processing by retry.

If the processing cannot be continued, the faulty disk device is immediately excluded from the processing, and the data transfer is continued by the remaining disk devices. Therefore, the processing time is also the delay time required for one retry. Occurs only.

However, if the processing can be continued by retrying, or if retries occur continuously, the delay time until the completion of data transfer increases, and it becomes impossible to complete the data transfer within the specified time.

In such a state, when the backup time of the backup power supply 9 elapses, it becomes impossible to supply the current, and it is impossible to save all the data stored in the semiconductor memory 3. There is a problem of data loss.

In view of such a problem, the present invention ensures that the data saving is completed within the backup capacity of the backup power supply 9 even when a failure occurs in the disk device and frequent retries occur. Has an aim.

[0023]

FIG. 1 is a block diagram for explaining the principle of the present invention. The semiconductor disk device is supplied with a current from the backup power supply 9 and stores data stored in the semiconductor memory 3 to which a plurality of logical drives are allocated,
Data is saved by sequentially reading out the data for each logical drive and simultaneously dividing and writing to a plurality of disk devices 6 to 8.

Then, the number of data error occurrences due to the failure occurrence of the disk devices 6 to 8 is calculated as follows.
When the data saving process is continued by the table 20 for recording the data for each disk device and the flag for instructing the disconnection of each of the disk devices 6-8 and the retry due to the failure occurrence of the disk devices 6-8, The time required for the data saving process is counted for each drive, and when the total of the counted time exceeds a predetermined time, the table 20 is referred to and the disk device having the largest number of data error occurrences is disconnected. The control means 21 which performs is provided.

Then, the predetermined time is set so that the data saving processing time required after the disconnection processing of the disk device is added to the predetermined time to be a predetermined time or less determined by the capacity of the backup power supply 9. When the flag is recorded in the table 20, the number of times the data error occurs that is set and the data saving process is performed is recorded in the table 20, and the disk device designated by the flag is excluded. Then, the data saving process is continued.

[0026]

With the above-described configuration, the recovery device can be optimized because the disk device in which the failure has occurred is disconnected after the recovery process has been executed up to the optimal number of retries for the capacity of the backup power supply 9. In addition to being able to do so, the data saving can be surely completed.

[0027]

2 is a block diagram of a circuit showing an embodiment of the present invention, FIG. 3 is a diagram for explaining an example of a table, and FIG. 4 is a diagram for explaining a data transfer time monitoring method.

In FIG. 2, the same reference numerals as those in FIG. 8 indicate the same functions. The processor 22 has a different program from the processor 17, and uses the table 20 and the timer 23 to control the disconnection of the disk device in which the number of data error occurrences is large and to count the data save time.

The table 20 has a structure as shown in FIG. 3, and the table A has disk devices 6 to 6 as shown in FIG. 3 (A).
8 and the disconnection flag for instructing disconnection are recorded.
As shown in (B), recording of a completion flag indicating whether or not data transfer is completed for each logical drive number, recording of the transfer capacity of the logical drive, recording of the standard time required for data transfer, and data transfer The actual time required to do this is recorded and recorded.

The processor 22 refers to the contents of the table 20 and monitors the data transfer time by the method shown in FIG. That is, when the time is plotted on the vertical axis and the transferred data amount is plotted on the horizontal axis, the transferred data amount at the time when the data transfer is completed is X in the column shown in the total of FIG. 3B.

The total standard processing time is shown in FIG.
It is Y shown in the total of (B), and the elapsed time during processing is a straight line.
It is indicated by (d). The straight line (a) is a specified time monitored by the control monitoring unit 4 shown in FIG. 7, that is, a time determined by the backup capacity of the backup power supply 9.

Further, the straight line (b) is the time monitored by the processor 22, and the straight line (c) shows the remaining time until the data transfer is completed after disconnecting the disk device in which the failure has occurred, and has a margin. When the amount of transferred data is 0,
It is set larger than the time indicated by Y.

Therefore, the transferred data amount of the straight line (b) is 0.
The time when is the time when the transferred data amount of the straight line (b) is X and the time when the transferred data amount of the straight line (c) is 0 is the straight line (a) It is set so that it does not exceed and is parallel to the straight line (d).

The straight line (e) shows the time actually required for data transfer when a failure occurs in the disk device.
Represents the course of Z shown in the total. When the processor 22 is instructed by the control monitor unit 4 shown in FIG. 7 to perform the data saving operation, the processor 22 sets the mode of the data managers 11 to 13 via the control circuit 18, and then activates the timer 23 to start the data manager 11 To 13 to start data transfer.

When the data transfer of one logical drive is completed, the data managers 11 to 13 are checked to check the error occurrence status and the timer 23.
Read the transfer time from.

Then, as shown in FIG. 3A of the table 20, for example, if an error occurs once in the disk device 6, the error occurrence frequency (1) is recorded and the error is recorded in the disk device 7 as 15 times. If it has occurred once, the number of error occurrences
If (15) is recorded and no error occurs in the disk device 8, the number of error occurrences (0) is recorded.

Logical drive number (0) shown in FIG. 3 (B)
When the data transfer is completed, the completion flag "1" is recorded, and if the time indicated by the timer 23 is 1 minute, the execution time
Record as (1).

If no failure occurs in the disk devices 6 to 8, the data transfer of the data of all the logical drives is completed in the standard time shown in FIG. 3B, and the total time becomes Y, so the processor 22 Does not exceed the monitoring time, that is, the straight line (b) shown in FIG.

If, for example, a failure occurs in the disk device 7 and retry processing occurs frequently, the logical drive number
If the data transfer time of (1) is 1 minute as standard,
When it takes time to complete and exceeds the time monitored by the processor 22, that is, the straight line (b) shown in FIG. 4, the processor 22 refers to the table 20 and the disk device 7 operates as shown in FIG. Since the maximum number of error occurrences is shown, "1" is set in the disconnection flag column of the disk device 7 to instruct the data manager 12 to disconnect the disk device 7.

Then, the disk drives 6 and 8 are used to execute the data transfer processing of the remaining logical drives (2) to N. 5 and 6 are flowcharts for explaining the operation of FIG.

Upon receiving the data transfer start in step (1), the processor 22 stores the contents of the table 20 in step (2). That is, the transfer capacity and standard time are stored in correspondence with the logical drive number.

Then, the transfer mode of the data manager or the like is set in step (3), the timer 23 is started in step (4), and it is checked in step (5) whether the disconnection flag is set in the table 20.

If the separation flag is not set, data transfer is started in step (6). Then, in step (7), it is checked whether an error has occurred.If no error has occurred, it is checked in step (11) if the monitoring time has expired. If not, in step (12) whether one logical drive transfer has ended. Find out.

If the transfer of one logical drive is not completed, the process returns to step (7), and if it is completed,
In step (13), the timer 23 is reset and the table 20
Update the contents of. That is, the completion flag is set and the execution time is recorded.

Then, in step (14), it is checked whether the transfer of all the logical drives is completed, and if not completed, the process returns to step (3), and if completed, it is judged as a normal end. If an error has occurred in step (7), the record of the number of error occurrences in the table 20 is updated in step (8). Then, in step (9), it is checked whether retry processing is possible, and if possible, retry is executed in step (10), and step (1
Move to the process of 1) and if retry is not possible, proceed to step (1
In step 6), set the disconnection flag of table 20 and
In step 7), the processing is continued using the remaining disk device, and the process returns to step (5).

When the monitoring time is exceeded in step (11),
In step (15), the number of error occurrences in the table 20 is referred to, and the process proceeds to step (16). If the disconnect flag of the table 20 is set in step (5), it is checked in step (18) whether the data is new data transfer. If data transfer of one logical drive is in progress, check whether the transfer of one logical drive is completed in step (19), and if it is completed, step (13)
If not completed, the data transfer is continued in step (21), and it is checked in step (22) whether an error has occurred.

If no error occurs, the process returns to step (19). If an error occurs, it is checked in step (23) whether retry processing is possible. If the retry process is possible, the retry process is executed in step (24), the process returns to step (19), and if the retry process is not possible in step (23), the process ends abnormally.

If it is new data transfer in step (18), data transfer is started using the remaining disk devices in step (20), and the process proceeds to step (21).

[0049]

As described above, according to the present invention, when a failure occurs in the disk device for saving data, the delay time and the time required for the remaining data transfer are managed to control the disconnection timing of the failed disk device. Therefore, the rescue process can be performed as much as possible, and the data saving process can be completed within the specified time.

[Brief description of drawings]

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

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

FIG. 3 is a diagram illustrating an example of a table.

FIG. 4 is a diagram illustrating a method of monitoring a data transfer time.

FIG. 5 is a flowchart explaining the operation of FIG. 2 (No. 1).

FIG. 6 is a flowchart (part 2) explaining the operation of FIG. 2;

FIG. 7 is a block diagram illustrating an example of a conventional technique.

FIG. 8 is a detailed block diagram of a disk control unit.

[Explanation of symbols]

 1 interface control unit 2 memory access control unit 3 semiconductor memory 4 control monitoring unit 5 disk control unit 6 to 8 disk device 9 backup power supply 10 data restoration circuit 11 to 13 data manager 14 to 16 buffer memory 17, 22 processor 18 control circuit 19 Data bus 20 Table 21 Control means 23 Timer

Claims (1)

[Claims] 1. A semiconductor memory (3) having a plurality of logical drives allocated thereto, which is supplied with a current from a backup power supply (9).
A semiconductor disk device that performs data save processing by sequentially reading the data stored in each logical drive, and simultaneously dividing and writing to a plurality of disk devices (6) to (8), The number of data error occurrences due to the failure occurrence of the disk devices (6) to (8) is recorded for each of the disk devices (6) to (8), and the disconnection of each of the disk devices (6) to (8) is performed. If the data saving process continues due to the table (20) recording the instructing flag and the retry due to the failure of the disk device (6) to (8), the time required for the data saving process for each logical drive. And counting means, when the total of the counting time exceeds a predetermined time, referring to the table (20), a control means (21) for performing a process of disconnecting the disk device with the maximum number of data error occurrences, After the disconnection process of the disk device Execute the data saving process by setting the required data saving process time so that the time added to the specified time is less than or equal to a certain time determined by the capacity of the backup power supply (9). The number of data error occurrences that occur during recording is recorded in the table (20), and the flag is set in the table (20).
When the data is recorded in, the semiconductor disk device is characterized in that the disk device specified by the flag is excluded and the data saving process is continued.