JP2001075741A - Disk control system and data maintenance method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the writing performance into a disk device and the maintainability of data. SOLUTION: Written data are stored in a log memory 171 built in a raid booster card (RAID BOOSTER) 17, and when data for one parity block are prepared, the data are collectively written in a raid disk array. The written data are stored in the log memory 171 of the card 17 and its copy also is stored in a main memory 13 of this computer system. When a fault is detected in the memory 171 or the card 17, the copy of the written data stored in the main memory 13 is quickly written in the disk array.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk control system and a data preservation method used in a computer system, and more particularly, to a disk control system having a function of collectively writing write data stored in a log memory to a disk device and its data. Conservation methods.

[0002]

2. Description of the Related Art In recent years, RAID (Redundant Array) has been developed from the viewpoint of speeding up disks and improving reliability.
PC servers equipped with technology of Independent Disks) have become mainstream. RAID technology includes types such as 0, 1, 10, and 5. Among them, RAID 5 is widely used because of its good price / capacity ratio and excellent fault tolerance. RAID5 is composed of a disk array in which a parity group including data and parity is arranged across a plurality of disk devices. R
The disk array of AID5 includes N + 1 (N is 2 or more) disk devices in which one disk device for parity storage is added to a plurality of disk devices for data storage, and parity is N + 1 (N + 1) in stripe units. N is 2 or more) so that the concentration of accesses to the parity disk can be prevented by distributing the disk devices to two or more disk devices.

However, when updating data in RAID 5, the following processing is performed in the RAID controller.

(1) Read old data and old parity data (2) Add new data and calculate new parity (3) Write new data and new parity data Among them, read old data and read old parity data Reading and writing of new parity data are essentially unnecessary processes and are called RAID5 write penalties. After all, in RAID 5, write performance is sacrificed in exchange for improvement in reliability. Therefore, RAI
In the PC server using D5, the writing performance (particularly, random writing) is poor, and it is necessary to construct a system that is aware of this. Although the latest RAID controller has a mechanism for reducing the write penalty, a great improvement cannot be expected for random write.

[0005]

Therefore, recently, R
As a technique for greatly improving the AID writing performance, a log structured file system technique is considered. The log-structured file system technology is a technique for improving performance utilizing the characteristics of a disk that "sequential access is much faster than random access". The performance degradation at the time of random access is caused by the seek and rotation wait of the disk, and is a part that should be eliminated if possible. Therefore, the log structured file system technology uses a log memory to store the write data, and writes the write data of a plurality of small blocks.
A technique of converting the data into data of one large block and performing “collective writing” is used.

However, when the log structured file system technology is used, the write data is "collectively written" to the disk. Therefore, there is a time when the write completion is reported to the application and a time when the actual write to the disk is performed. different. For this reason, write data that should have been already recorded on the disk remains only in the log memory, so that sufficient consideration must be given to the integrity of the data.

The present invention has been made in view of such circumstances, and provides a disk control system and a data maintenance method capable of realizing an improvement in write performance to a disk device and an improvement in data integrity. The purpose is to:

[0008]

In order to solve the above-mentioned problems, the present invention provides a disk control system for controlling a disk device provided in a computer system, which has a log memory for storing write data. Control means for writing the write data stored in the log memory to the disk device at a time; and means for holding a copy of the write data stored in the log memory. The duplication is used to ensure the integrity of the write data in the log memory.

In this disk control system, the write data can be duplicated by holding not only the log data but also a copy of the write data in the memory of the computer system. For this reason, even when an error occurs in which normal data input / output to / from the log memory is not performed, for example, when an error is detected, a process of writing a copy of the write data to the disk device is automatically performed. The original contents of the write data can be restored using the copy.
Therefore, it is possible to improve the write performance to the disk device and the data integrity. In particular, by applying to the control of a disk array in which a parity group including data and parity is traversed with respect to a plurality of disk devices, it is possible to improve write performance and reliability at a high level. Become.

The detection of the error in the log memory is preferably performed by providing a status memory for holding error status information indicating an error in the log memory in the control means and periodically checking the status memory. Usually, in a computer system, when a memory error or the like occurs, a reset signal is automatically issued, so that the computer system itself is often stopped.
However, by using the error detection method by polling as described above, errors in the log memory can be confined in the control means, and data recovery processing using duplication of write data can be performed efficiently. Becomes

According to the present invention, there is provided a disk control system for controlling a disk device provided in a computer system, comprising a log memory for storing write data, and storing the write data stored in the log memory. Control means for writing to the disk device in a lump, error detecting means for detecting an error of the disk device, and writing data stored in the log memory in response to error detection by the error detecting means. An evacuation means for ensuring the integrity of the write data stored in the log memory using the saved write data.

In this computer system, if batch writing is not performed due to an error in the disk device, the write data stored in the log memory is saved, whereby the data is written out to the disk device. The contents of the write data remaining only in the log memory are automatically backed up. Therefore, even if a disk failure occurs, it is possible to restore the contents of the write data, so that the time at which the write completion is reported to the application is different from the time at which the actual write to the disk is performed in the system. It is possible to improve the maintainability.

[0013]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a configuration of a computer system according to an embodiment of the present invention. This computer system is used as a server (PC server).
A plurality of CPUs 11 can be mounted. Each of these C
The PU 11 is connected to the bridge 12 via the processor bus 1 as shown. The bridge 12 is a bridge LSI for connecting the processor bus 1 and the PCI bus 2 in two directions, and has a built-in memory controller for controlling the main memory 13. Main memory 13
Is loaded with an operating system, an application program to be executed, a driver, and the like.

As shown, the PCI bus 2 has an SCS
I controller 14, RAID controller 16, and RAID booster card (RAID BOOSTE
R) 17 is connected. SCSI controller 14
(HDD) 15 is a system disk used to record an operating system and the like. The disk array 18 controlled by the RAID controller 16 is used for recording various user data and the like.

The disk array 18 functions as a RAID 5 disk array under the control of the RAID controller 16. This disk array 18 has N + 1 disks (here, DISK0 to DIK) in which one disk device for parity storage is added to N disk devices for data storage.
SK4). These N + 1 disk devices are grouped and used as a single logical drive.

As shown, data (DATA) and its parity (PAR) are stored in the grouped disk devices.
ITY), and the parity position is moved to N + 1 disk devices for each stripe. That is, in the stripe S0, the parity (PARITY) of the data (DATA) group on the stripe unit assigned to the same position of each of DISK0 to DISK3 is DI
It is recorded on the corresponding stripe unit of SK4, but in the next stripe S1, that stripe S
The parity (PARITY) corresponding to 1 data is DI
It is recorded on the corresponding stripe unit of SK3.
By dispersing the parity in the stripe units among the N + 1 disk devices in this way, it is possible to prevent the concentration of accesses to the parity disk.

RAID booster card (RAID BO
The OSTER 17 is for improving the write performance to the RAID 5 disk array 18 by using the log structured file system technology, and is realized as a PCI expansion card that can be mounted in a PCI slot. Here, referring to FIG.
The principle of the write control by the booster card (RAID BOOSTER) 17 will be described.

In the writing method using the log structured file system technology, the data writing position is not determined according to the logical address requested by the host to write, but by storing the writing data in the order of writing requested by the host. A large data block for writing is formed, and the large data block is collectively written in a free area of the disk array 18 from the top in order. This makes it possible to convert random access into sequential access. Here, the unit of “collective writing” is the data capacity of one parity group, and the write start position is aligned with the head of the parity group, thereby making the RAID controller 1
In No. 6, parity can be calculated only from the write data of “collective writing”. Therefore, the above-described reading of old data, reading of old parity data, and writing of new parity data (write penalty) do not occur, and the write performance is improved.

FIG. 2 shows that the block data size of the write data from the application program sent via the file system is 2 KB, the data size of one stripe unit is 64 KB, and the data size of one stripe (parity group) is 256 KB ( = 64 KB
・ This is an example of the case of 4). 256KB of data is RA
ID Booster Card (RAID BOOSTER) 17
At this time, writing to the disk array 18 is performed at once.

RAID booster card (RAID BO)
OSTER) 17, as shown in FIG. 1, a log memory 171, an address mapping table 172,
ECC status register 173, DMA engine 17
4, MPU 175, battery 176, and the like.

The log memory 171 is for accumulating write data from an application, and when write data for one parity group is accumulated in the log memory 171, collective writing to the disk array 18 is performed. Address mapping table 172
Is the logical address of the write data and the disk array 1
8 is used to manage the correspondence with the actual write position. That is, the correspondence between the value of the logical address requested to be written and the actually written physical address is managed by the address mapping table 172.

The log memory 171 and the address mapping table 172 are stored in a RAID booster card (RAID B).
OOSTER) 17 is implemented on a memory mounted thereon, and the contents of the memory are backed up by a battery 176.

The ECC status register 173 is an error status register for holding error status information indicating an error in the log memory 171.
11 can be read. The DMA engine 174 is used for performing data transfer between the log memory 171 and the main memory 12 by DMA transfer. MPU175 is a RAID booster card (RAID
BOOSTER) 17 is used to control each unit.

FIG. 3 shows a RAID booster card (RA
The relationship between hardware and software for performing disk control using the ID BOOSTER 17 is shown.

The disk control system according to the present embodiment has an R
AID Booster Card (RAIDBOOSTER) 17
And control software. The control software includes a filter driver 23 for performing a log structuring process and a RAID booster card (RAID BO).
OSTER) 17 for controlling the card driver 25
It is composed of

The filter driver 23 is located between the file system 22 of the OS and the RAID controller driver 24, and the write data from the application program 21 is automatically acquired by the filter driver 23. The acquired write data is accumulated in the log memory 171 through the card driver 25. When data for one parity group is accumulated in the log memory 171, under the control of the filter driver 23,
Write data is transferred to the RAI controller 16 through the RAID controller driver 24.

(Improvement of the integrity of write data)
Next, the data security function of the present embodiment will be described.

As described above, the point of the performance improvement by the RAID booster card (RAID BOOSTER) 17 is to collect write data from the application and collectively write the data to the disk array 18 collectively. Therefore, since write data which should be originally recorded in the disk array 18 remains only in the log memory 171, sufficient consideration must be given to the integrity of the data. Therefore, in this embodiment,
The following measures are taken.

(1) Backup of Memory by Battery Battery module 176 is connected to RAID booster card 1
7 so that the contents of the memory can be backed up for 72 hours (during normal use) with a battery backup even in the event of a power failure or the like.

(2) Data Protection by ECC The memory on the RAID booster card 17 is protected by ECC, so that even if a 1-bit error occurs in the memory, it can be repaired and data is not lost.

(3) Duplication of Write Data Only with the above measures, R
Due to the failure of the AID booster card 17 itself, the write data stored in the log memory 171 is lost. Therefore, the following processing is performed on the write data to further improve data integrity.

(A) The write data is copied not only in the log memory 171 of the RAID booster card 17 but also in the main memory 13 of the computer system.
A copy of the write data is recorded on a work area secured on the main memory 13 as a log memory copy area.

(B) In a normal PCI board, when a failure occurs, a reset signal is generated and the server is stopped (fail stop). However, in the RAID booster card 17, even if a failure occurs, the influence of the failure is controlled so as not to be confined to the card and spread to the computer main body (fault isolation), so that the software in the computer can be directly executed. . For this purpose, the aforementioned ECC status register 173 is prepared so that the state of the log memory 171 can be monitored from software.

(C) The control software on the computer always checks the status of the RAID booster card 17 by using the ECC status register 173, and the log memory 171
When a failure of the card or the card is detected, a copy of the write data on the main memory 13 is immediately copied to the disk array 18.
Write to

With this measure, even if the RAID booster card 17 fails, the write data will not be lost if the control software on the computer side is operating at that time.

(4) Dump File at Disk Failure In this embodiment, the write data is "collectively written" to the disk array 18, so that the writing completion is reported to the application and the writing to the disk array 18 is actually performed. The time is different. Therefore, if a disk failure occurs during writing to the disk array 18,
Data to be written to the disk array 18 remains on the log memory 171.

In the event of a disk failure, the stored write data is stored as a dump file in the system disk 15 or the like, and when the disk failure is recovered, the dump file is automatically stored in the disk array 1.
8, data consistency as seen from the application is ensured.

(Duplicate of Write Data) Next, with reference to FIG. 4 to FIG. 6, the principle of processing by duplicating write data will be described. Here, for simplicity of description, it is assumed that the disk array 18 is composed of N + 1 (N = 3) disks, and that the block data size of one stripe unit and the write data from the application are equal.

As shown in FIG. 4, the write data from the application is stored in the log memory 171 of the RAID booster card (RAID BOOSTER) 17.
And a copy thereof is created on the log memory copy area 131 of the main memory 13. Then, as the operation of the system proceeds, the write data is accumulated in the log memory 171, and when the write data for one parity group is accumulated in the log memory 171, FIG.
As shown in (1), collective writing to the disk array 18 is performed.

Before the write data for one parity group is prepared in the log memory 171, the log memory 171
When the failure of the RAID booster card 17 including the error described in (1) occurs, the write data recorded in the log memory 171 up to that time cannot be used. In this case, as shown in FIG.
Writing to the disk array 18 is performed using the copy data recorded above. At this time, write data for one parity group is created by adding arbitrary data as dummy data to the copy data, and the data is "collectively written". As a result, the RAID booster card 1 including the error in the log memory 171
7, the RAID controller 16
Can calculate the parity only from the write data of “collective writing”.

(Data Write Processing) Next, the procedure of the data write processing in this embodiment will be described with reference to the flowchart of FIG.

When obtaining write data from the application program 21 via the file system 22 of the OS, the filter driver 23 first activates the card driver 25 and stores the write data in the log memory 1.
A request is made to the card driver 25 to execute a DMA transfer for writing to the memory 71 (step S101). Next, the filter driver 23
Is copied to the log memory copy area 131 allocated on the work area of the filter driver 23 or the like, and a copy of the write data is created (step S102). Thereafter, the filter driver 23 waits for a predetermined time until a DMA completion notification is returned from the card driver 25 (step S10).
3) It is determined whether the RAID booster card (RAID BOOSTER) 17 including the log memory 171 is normal (step S104).

In this case, for example, due to a failure of the log memory 171 or the RAID booster card 17 itself, the DM
If the writing by A is not completed normally and the DMA completion notification is not returned even after the lapse of a predetermined time, it is determined that some failure (log memory error) in which data cannot be written to the log memory 171 has occurred. Is determined. Even if the DMA completion notification is returned within a predetermined time, if an error status indicating the occurrence of a 2-bit error is set by referring to the ECC status register 173, a log memory error will be generated. It is determined that it has occurred.

When such a log memory error has not occurred, the RAID booster card (RAID BOOST
ER) 17 is operating normally (step S1)
04, YES), the filter driver 23 reports the completion of writing to the application program 21 (step S105). Then, the filter driver 23 determines whether or not the write data for one parity group has been collected in the log memory 171 (step S10).
6) If all the data are collected, the batch write processing of the parity group is executed, and the write data of one parity group is "collectively written" to the disk array 18 and the processing is terminated (step S107). If the write data for one parity group is not available in the log memory 171, the process ends without executing step S107.

On the other hand, when occurrence of a log memory error is detected, that is, when a RAID booster card (RAID B
If the OOSTER) 17 is not operating normally (NO in step S104), the filter driver 23
By adding dummy data to the copy data already existing in the log memory copy area 131 before the execution of step S102, write data for one parity group is created (step S108). And
The filter driver 23 executes a parity group batch write process, and writes the write data for one parity group created in step S108 to the disk array 1.
After the “collective writing” in step 8 (step S109), the application program 21 is notified of the occurrence of a write error, and the process ends (step S110).

As described above, a failure is detected each time write data is recorded in the log memory 171, and when a failure is detected, the copy data is immediately transferred to the disk array 1.
By writing an 8 and returning an error to the host system requesting the writing, it is possible to improve the integrity of the write data.

(Data Read Processing) Next, the procedure of the data read processing in the present embodiment will be described with reference to the flowchart of FIG.

Upon receiving a data read request from the application program 21 via the OS file system 22, the filter driver 23 first determines whether or not the read requested data exists in the log memory 171. (Step S201). This determination is made by checking whether there is any write data that matches the logical address requested to be read among the write data recorded in the log memory 171 in response to the write request.

The data requested to be read is stored in the log memory 17.
If not present (step S201: NO), the filter driver 23 controls the RAID controller 16 via the RAID controller driver 24 and reads the corresponding data from the disk array 18 (step S206). In this case, since data is sequentially written to the disk array 18 in the order of write requests, the physical address from which to read data is determined by the filter driver 23 by referring to the address mapping table 172. After reading the data, the filter driver 23 notifies the application program 21 of the read completion, and ends the processing (step S207).

The data requested to be read is stored in the log memory 17.
1 (YES in step S201),
The filter driver 23 activates the card driver 25 to read the data requested to be read.
It requests the card driver 25 to execute the transfer (step S202). Then, the filter driver 23 waits for a predetermined time for the DMA completion notification to be returned from the card driver 25 (step S203), and then returns to the log memory 171.
RAID Booster Card (RAID BOOST)
ER) 17 is determined to be normal (step S)
204).

For example, the log memory 171 or RAID
Due to the failure of the booster card 17 itself, the reading by the DMA is not completed normally and the DMA
If no completion notification is returned, the log memory 171
It is determined that some failure (log memory error) in which data cannot be read from the server has occurred. In step S204, the ECC status register 173 is also checked. If an error status indicating the occurrence of a 2-bit error is set, it is determined that a log memory error has occurred.

When such a log memory error has not occurred, the RAID booster card (RAID BOOST
ER) 17 is operating normally (step S2)
04 (YES), the filter driver 23 reports the read completion to the application program 21 and ends the processing (step S205).

When the occurrence of a log memory error is detected, that is, when a RAID booster card (RAID BOO
If the (STER) 17 is not operating normally (NO in step S204), the filter driver 23 adds dummy data to the copy data existing in the log memory copy area 131, so that one parity group is added. Is created (step S20).
8). Then, the filter driver 23 executes a batch write process of the parity group, and executes step S208.
After the write data for one parity group created at step S is “collectively written” to the disk array 18 (step S
209), the occurrence of a write error is notified to the application program 21, and the process is terminated (step S210).

As described above, the failure is detected even at the time of data reading, and when the failure is detected, the copy data is immediately written to the disk array 18, thereby improving the integrity of the write data. It becomes possible.

(Dump File When Disk Failure Occurs) Next, processing for saving the contents of the log memory 171 as a dump file when a disk failure occurs will be described.
This save processing is performed in the “parity group batch write processing” described with reference to FIGS. 7 and 8.

FIG. 9 shows the procedure of the "parity group batch write process".

The filter driver 23 first sets the RAI
The D controller driver 24 is activated, and requests the RAID controller driver 24 to perform a DMA transfer for collectively writing data for one parity group to the disk array 18 (step S301). The RAID controller 16 automatically generates a parity from the data of one parity group, and tries to write the parity to the first stripe of the free area.

Then, the filter driver 23
After waiting for a predetermined time until the DMA completion notification is returned from the ID controller driver 24 (step S302), R
By examining the contents of the completion status from the AID controller driver 24, it is determined whether or not a failure has occurred in the disk array 18 (step S30).
3). If any failure has occurred in the disk array 18 and writing to the disk array 18 has failed (YES in step S303), the filter driver 23 sends the write data stored in the log memory 171 to the system disk 15 Is saved as a dump file (step S304). In this case, an identifier such as a sequence number is embedded at the beginning and end of the dump file as shown in FIG. 10 so that it can be checked whether or not the dump file has been correctly saved to the end.

(Data Restoring Process Using Dump File) Next, a data restoring process using a dump file will be described with reference to a flowchart of FIG.

That is, the computer system is restarted,
When the filter driver 23 is loaded, the filter driver 23 first determines whether a dump file exists on the system disk 15 and, if so, whether the contents of the dump file are correct. Processing is performed (step S401). This is because an incorrect dump file may be recorded, such as when the system goes down while writing the dump file.

Whether or not the contents of the dump file are correct can be determined by checking the above-mentioned sequence number. If a sequence number is recorded at both the beginning and end of the dump file, it is determined that the dump file is correct. In this case, the filter driver 23 executes a parity group batch write process, and writes data for one parity group in the dump file to the disk array 18 (step S402). If the data in the dump file is not enough for one parity group, the batch write processing may be executed after adding the dummy data.

As described above, in this embodiment, by providing functions such as duplication of write data and creation of a dump file in the event of a disk failure, the integrity of write data can be improved. Sufficient writing performance and sufficient reliability can be obtained. Also, RA
Although the function of the ID booster card 17 can be installed in the RAID controller, by implementing the RAID booster card 17 as a PCI card as in the present embodiment, the existing RAID controller can be changed without any change. In addition, it is possible to easily realize an improvement in write performance and an improvement in data integrity.

In this embodiment, the parity is generated by the RAID controller. However, the parity may be generated by the RAID booster card 17. The backup file may be stored anywhere as long as it is a non-volatile recording medium. further,
The data preservation method according to the present embodiment is particularly effective for a RAID 5 disk array, but applies a configuration in which “collective writing” is performed using the log memory 171 such as another RAID type disk array or a single disk device. It can be applied to any possible disk subsystem.

[0065]

As described above, according to the present invention,
By providing functions such as duplication of write data and creation of a dump file in the event of a disk failure, it is possible to achieve an improvement in write performance to a disk device and an improvement in data integrity.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a computer system according to an embodiment of the present invention.

FIG. 2 is an exemplary view for explaining the principle of “collective writing” performed by the system of the embodiment.

FIG. 3 is an exemplary view showing a relationship between hardware and software in the system according to the embodiment;

FIG. 4 is an exemplary view for explaining write data copy processing which is executed in the system of the embodiment.

FIG. 5 is an exemplary view for explaining batch write processing of log data executed in the system of the embodiment.

FIG. 6 is an exemplary view for explaining batch data write processing executed in the system according to the embodiment;

FIG. 7 is an exemplary flowchart illustrating a flow of a series of processes performed by the system according to the embodiment when writing data.

FIG. 8 is an exemplary flowchart showing the flow of a series of processes performed by the system of the embodiment when reading data.

FIG. 9 is an exemplary flowchart illustrating the procedure of a batch write process executed in the system of the embodiment.

FIG. 10 is an exemplary view showing an example of a dump file created when a disk failure occurs in the system of the embodiment.

FIG. 11 is an exemplary view showing an operation at the time of reboot executed in the system of the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 ... CPU 13 ... Main memory 15 ... System disk 16 ... RAID controller 17 ... RAID booster card 18 ... Disk array 23 ... Filter driver 25 ... Card driver 131 ... Log memory copy area 171 ... Log memory 173 ... ECC status register

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G11B 20/18 572 G11B 20/18 572F

Claims (11)

[Claims] 1. A disk control system for controlling a disk device provided in a computer system, comprising: a log memory for storing write data; and collectively storing the write data stored in the log memory. Control means for writing to the disk device; and means for holding a copy of the write data stored in the log memory, and using the copy of the write data, to maintain the integrity of the write data in the log memory. A disk control system characterized by securing. 2. The apparatus according to claim 1, further comprising: an error detecting unit that detects an error in the log memory; and a unit that writes a copy of the write data to the disk device in response to the error detection by the error detecting unit. The disk control system according to claim 1, wherein 3. The control unit has a status memory that holds error status information indicating an error of the log memory. The error detection unit periodically checks the status memory to check the log memory. 3. The disk control system according to claim 2, wherein presence or absence of an error is detected. 4. The disk device comprises a disk array in which a parity group including data and parity is arranged across a plurality of disk devices, and the control means stores one parity in the log memory. 2. The disk control system according to claim 1, wherein writing to the disk array is performed in units of the one parity group each time write data for a group is prepared. 5. A disk control system for controlling a disk device provided in a computer system, comprising a log memory for storing write data, and collectively storing the write data stored in the log memory. A control unit for writing to the disk device; an error detection unit for detecting an error of the disk device; and a unit for saving write data stored in the log memory in response to the error detection by the error detection unit. A disk control system comprising: using the saved write data to ensure the integrity of the write data stored in the log memory. 6. The disk control system according to claim 5, further comprising a data restoring unit that writes the saved write data to the disk device when the computer system is restarted. 7. The system according to claim 1, further comprising: a unit for judging the validity of the saved write data when the computer system is restarted. 7. The disk control system according to claim 6, wherein the disk control system is executed when the performance is confirmed. 8. A data security method applied to a disk control system for controlling a disk device provided in a computer system, wherein write data is stored in a log memory and a copy of the write data is held. And collectively writing the write data stored in the log memory to the disk device, and using the copy of the write data to ensure the integrity of the write data in the log memory. A data preservation method characterized by the following. 9. The method according to claim 1, further comprising: an error detection step of detecting an error of the log memory; and a step of writing a copy of the write data to the disk device in response to the error detection of the log memory. 9. The data security method according to claim 8, wherein 10. A data security method applied to a disk control system for controlling a disk device provided in a computer system, comprising: a step of storing write data in a log memory; and a step of storing write data in the log memory. Collectively writing the write data to the disk device; detecting an error in the disk device; and evacuating the write data stored in the log memory in response to the error detection in the disk device. Using the saved write data to secure the integrity of the write data stored in the log memory. 11. The method according to claim 1, further comprising the step of writing the saved write data to the disk device when the computer system is restarted.
0 Data security method.