JPH0318216B2 - - Google Patents

Description

Detailed Description of the Invention (1) Technical Field of the Invention The present invention relates to a journal creation and management method in a data processing system.

(2) Technical background Magnetic disk drives are capable of rapid writing and reading;
It is often used as a file because it has characteristics such as non-volatility and large storage capacity, but as a countermeasure in case an error occurs in the stored contents, a journal file is prepared in addition to the general file and written to the general file. The data is also written to the journal file. Figure 1 shows an overview of this type of file system, where 1 is the host
A CPU (central processing unit), 2 is a control device for the file device 5, 6 is a general file, 7 is a journal file, and 5A and 5B are drive units for the magnetic disk device. Update data of the general file is written to the journal file 7, and if there is an error in the update data read from the general file, the update data stored in the journal file is read and the general file is updated again to obtain correct data. do it like this. Since the capacity of the journal file is limited, it is not possible to copy all the data in the general file to the journal file, but all recently updated data is in the journal file, and it is possible to restore the general file for the recently updated data. do.

(3) Prior art and problems When the host CPU 1 issues a write instruction to any one of multiple files 6 and updates the file, in order to take the journal,
CPU 1 must instruct the appropriate track of journal file 7 to write the same data. To do this, commands for the journal file must be reconfigured and the journal file must also be managed, which increases access to the control unit 2, increases the load on the host CPU, and reduces system efficiency. Deteriorate.

Nowadays, control devices that are capable of storing the latest data inside the control device have become available, and in the case of magnetic disks, this is called a disk cache control device. By using this data storage function, you can collect data without increasing the load on the host CPU.

(4) Purpose of the Invention The present invention focuses on such points, and provides a control device with journal creation and management functions,
This is an attempt to reduce the burden on hosts.

(5) Structure of the Invention The present invention provides a journal creation management method in a data processing system comprising a general file and a journal file for recording update data to the general file, a control device for the file device, and a host CPU. The control device is provided with a data buffer in which updated data for general files is stored and a control table that stores management information for the data buffer and journal file, and when updating the general file, the host CPU sends the updated data to the control device. Instructing to update a general file, the control device updates the general file, simultaneously writes the updated data to its own data buffer, writes control data including a flag to the control table, and performs a series of writes to this general file. After the operation is completed, the control device refers to the flags set in the control table, and if there is a flag indicating unprocessed journal, the data buffer is added to the corresponding track of the journal file according to the control data to which the flag belongs. This is characterized in that the update data of 1 is written.Next, this will be explained in detail with reference to embodiments.

(6) Embodiment of the invention FIG. 2 shows an embodiment of the invention. Although it is almost the same as Fig. 1, a disk cache control device is used as the control device, and therefore a large-capacity data buffer 4 is used.
There is. 3 is a control table, and data buffer 4
and various information for managing the journal file 7. Data buffer 4 is general file 6
The data on the track of the file is stored when the file is read/written. The general file 6 is a file that can be read/written by the system, and the journal file 7 is a file that stores written data when writing to the general file is performed as described above.

When the host CPU 1 gives an instruction to write to one of the files 6 (6a) to the control device 2, the control device 2 writes the data sent from the host CPU 1 to the corresponding track of the file 6a. At the same time, data is stored in buffer 4. Control table 3 when storing data in buffer 4
With reference to do.
Figure 3 schematically shows part of the contents of the control table (CT) 3 and data buffer (DB) 4. data),
FAI to FCI are pointers indicating where in the data buffer (in which track if the data buffer DB is a disk) information about files A, B, and C, for example, files A, B, and C are stored. After performing these series of operations, the write operation to the general file 6 is completed. After that, the control device 2 refers to the flags set in the control table 3 to search for a flag indicating that the journal has not been processed, and if the flag is found, data buffer is created in the corresponding track of the journal file 7 according to the control data to which the flag belongs. The data written in the file 6a in 4 is written. In this way, according to the method of the present invention, a journal file can be created using only the control device without imposing any load on the host, and the overhead of the host can be reduced.

When writing data from the journal file 7 to the general file 6, for example, the host CPU 1 instructs the control device 2 by a command as to which general file (assumed to be 6x) data is to be rewritten. In response to this, the control device 2 refers to the control table 3 and sets the journal file 7.
The corresponding portion of the file 6x is read and the read data is written to the corresponding portion of the file 6x. To transport data from this journal file to the general file, the host
Since the CPU is not involved, host overhead can be reduced when files are repaired, and operational errors can also be prevented when repairing files.

(7) Effects of the Invention As explained above, in the present invention, since the control device is equipped with an automatic journal function, the load on the host CPU can be reduced, and in the case of a control device with a built-in large-capacity buffer, even if the control device is equipped with a journal function, the load on the host CPU can be reduced. It is advantageous that the present invention can be carried out at low cost since there is no need to significantly modify the system.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an overview of an information processing system to which the present invention is applied, FIG. 2 is a block diagram showing an embodiment of the present invention, and FIG. 3 is a diagram explaining the contents of a control table and data buffer. . In the drawing, 6 is a general file, 7 is a journal file, 5 is a file device, 2 is a control device, 1 is a host CPU, 4 is a data buffer, and 3 is a control table.

Claims (1)

[Scope of Claims] 1. A journal creation management method in a data processing system comprising a file device having a general file and a journal file for recording update data to the general file, a control device for the file device, and a host CPU, comprising: A data buffer in which update data to the general file is stored and a control table to store management information for the data buffer and journal file are provided. When updating the general file, the host CPU sends the updated data to the control device and updates the general file. The control device updates the general file, simultaneously writes the updated data to its own data buffer, writes control data including flags to the control table, and executes a series of write operations to this general file. After completion, the control device refers to the flags set in the control table, and if there is a flag indicating unprocessed journal, the control device transfers the above data in the data buffer to the corresponding track of the journal file according to the control data to which the flag belongs. A journal creation management method characterized by writing updated data.