JPH02136922A - Sorting/merging process method for disk data - Google Patents

Abstract

PURPOSE:To reduce the shift frequency of a magnetic head between cylinders and also to shorten the seeking time by setting the magnetic head on one of both cylinders to continuously perform a comparing process for sorting and merging until just a single piece of data remains on the other cylinder then shifting the head to the other cylinder to perform the same comparing process. CONSTITUTION:A cylinder control part 1 gives an instruction to a buffer control part 2 to give an access to each cylinder via a magnetic head. At the outset (n - 1) pieces of data are read out of one of both cylinders and sent to (n) pieces of buffers together with a single piece of data read out of the other cylinder and sent to these buffers respectively. Then these (n) pieces of data are compared with each other for sorting and merging via a comparing process part 4. Thus the preferential data is outputted and the data on the cylinder where the magnetic head is set are fetched and compared repetitively with each other. When just a single piece of data remains on the other cylinder out of those (n) pieces of data, the magnetic head is shifted to the other cylinder to perform the same process again. Thus it is possible to reduce the shift frequency of the magnetic head between both cylinders and also to shorten the seeking time.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for classifying and merging disk data in a magnetic disk device as an external storage device, and more particularly to a method for classifying and merging disk data between cylinders.

[Conventional technology] Conventionally, this type of disk data classification/merging processing method was
In the process of reading data that spans between cylinders, the seventh
As shown in the figure, the general method is to read data alternately from each cylinder 5182 and classify and merge it, or to use the random access function unique to the disk to read data irregularly from each cylinder and classify and merge it. It became.

[Problem to be solved by the invention] The conventional disk data classification and merging processing method described above is
This method does not take into account the seek time required for positioning the magnetic head, which is unique to disk drives, and therefore has the disadvantage that it takes time to move the magnetic head each time data is loaded into or taken out from the buffer.

SUMMARY OF THE INVENTION An object of the present invention is to provide a disk data classification/merging processing method in which reading and writing of data are performed on the same cylinder as much as possible in order to minimize the seek time of a magnetic head peculiar to a disk device.

[Means for Solving the Problems] The disk data classification and merging processing method of the present invention calculates the number of data to be read initially on two different cylinders in which data is arranged in advance using keys for classification and merging. According to the cylinder control table display,
First, take n-1 pieces of data from one cylinder, and
Read one piece of data from the other cylinder into each of the n input buffers, and at the same time write the input buffer address and key for classification/merging of each of the n pieces of data read into the buffer control table, and input. The n pieces of data read into the buffer are compared using the keys in the buffer control table, and as a result, priority data is determined according to a certain order of keys and is input into the previously output blank space on the cylinder side where the magnetic head is currently located. The priority data is input from the buffer, and then the next data is read from the cylinder that accepted the priority data into the empty input buffer, and the number of data in the cylinder control table is read according to the processing content. and updates the contents of the buffer control table, and then checks the cylinder control table until the display changes from the initial data count ratio of the two cylinders, n-1 to 1, to l to n-1. Repeat the same process, then switch the magnetic head from the cylinder where the magnetic head is currently located to the other cylinder, and repeat the same data input/output processing and magnetic head switching to classify the data on the cylinder.・This will be repeated until the annexation is completed.

[Operation] First, according to the display of the cylinder control table, n-1 pieces of data are read from one cylinder and 1 piece of data from the other cylinder are read into n input buffers, and the buffers are loaded. These n pieces of data on the control table are compared using their respective keys for classification and merging. Therefore, data is determined in priority order according to a certain type of key, and data is written in the blank space after the previous input on the cylinder where the magnetic head is located at that time, and at the same time, the next data from the cylinder is written. Store it in the input buffer that was just output, and update the contents of the cylinder control table and buffer control table. By repeating this process, the cylinder control table display changes from the initial n-1:1 to 1:
When the rotation is reversed to n-1, the position of the magnetic head is switched from the current cylinder to the other cylinder and the same process is repeated. In other words, the present invention places a magnetic head on one cylinder, performs comparison processing for classification and merging until only one piece of data remains on the other cylinder, and at that point, places the magnetic head on the other cylinder. By moving the magnetic head and performing the same processing, seek time can be saved by reducing the number of times the magnetic head moves between cylinders as much as possible.

[Example] Next, an example of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of the disk data classification/merging processing method of the present invention, FIG. 2(a),
(b) is a configuration diagram of the cylinder control table and buffer control table used in this embodiment, FIG. 3 is a flowchart showing the processing procedure according to this embodiment, and FIG. 4 is a block diagram of cylinders S1 and St in the case of merging processing. Figures 5(a), 5(b), and 5(c) show examples of the arrangement of the above initial read data and written data after merging processing.
are diagrams showing the progress between the contents of input buffers A, B, C, D, and H and the display of the cylinder control table and the buffer control table, respectively. FIG. 6 is a diagram showing the progress of the writing operation of the magnetic head in the case of FIG. 4. , cylinders S1 and 8
It is a figure showing movement between two.

The cylinder control table consists of a column 6 for writing the number n-1 of data to be read from one cylinder S1 into the manual buffer, and a column 7 for writing the number 1 of data to be read from the other cylinder S2 to the input buffer. The contents are updated every time the processing of a step is completed. In addition, the buffer control table has a column 8 in which a pointer to the cylinder control table is written, and a manual buffer address and a key for classifying and merging data for data on the cylinder stored in input buffers A to H are written. n-
It consists of one entry field 9, and two entries are created corresponding to each cylinder. All of these tables are stored in main memory (not shown). The cylinder control section 1 refers to the cylinder control table and causes the buffer control section 2 to access each cylinder by the magnetic head according to the display, and also checks the display of the cylinder control table each time to position the magnetic head. Instructs switching between cylinders so that the The buffer control unit 2 controls the input processing unit 3 according to instructions from the cylinder control unit 1.
reads data from the cylinders to input buffers A-E via the comparison processing unit 4, compares these data with keys for classification and merging according to the buffer control table, and determines a winner based on certain keys; The output processing unit 5 outputs this from the input buffer, and inputs the next data on the cylinder after the empty space.

Next, the operation of this embodiment will be explained with reference to the flowchart of FIG.

In the cylinder control table, initially, the number n− of data to be read from one cylinder (hereinafter referred to as cylinder St) is
1 and the number of data to be read from the other cylinder (hereinafter referred to as cylinder S2), 1, are entered. At the start of the operation, the cylinder control unit 1 instructs the buffer control unit 2 to take in n-1 pieces of data from the cylinder S1 and 1 piece of data from the cylinder S2, and the buffer control unit 2 instructs the input processing unit 3 to take in n-1 pieces of data from the cylinder S1 and 1 piece of data from the cylinder S2. data is taken in, and its contents are entered in each buffer control table (step 31). After writing, the magnetic head is located on the cylinder S2. Next, the buffer control unit 2 compares each piece of data in the buffer control table using its key, and determines the data to be selected as the "winner" according to a certain standard (step 32), and selects this winner as the "winner". The next data on cylinder S2 is read out first and output to an empty location on cylinder S2 (step 33), and the next data on cylinder S2 is read into the empty manual buffer after that, and the contents of the buffer control table are updated. In response to the notification from the buffer control unit 2, the cylinder control unit 1 updates the contents of each column 6 and 7 of the cylinder control table according to the processing result (
Step 34). Next, the cylinder control unit 1 refers to the cylinder control table and determines that the ratio of the data numbers in columns 6 and 7 is 1.
: Check whether it has become n-1 (step 3)
5) If not, it is checked whether the merging or classification has been completed (step 37), and if it has not been completed, the same operation is repeated from step 32. Also,
When the contents of columns 6 and 7 of the cylinder control table become 1;n-1 in step 35, the cylinder control unit 1 switches the magnetic head from cylinder S2 to cylinder S1 (step 36), and ends in step 37. Unless otherwise, the same operation is repeated from step 32.

In this way, the operation ends with the completion of the merging and classification work (step 37).

Next, the above-mentioned operation will be explained with reference to FIG. 4, FIG. 5, and FIG. 6 for a more specific case of merging.

The input data shown in the left column of cylinder S or Sz in FIG. 4 is stored on cylinders Sl and S2 of the disk in order according to their respective classification/merging keys (numbers). At the start of the process, the cylinder control unit 1 first sends the fifth data via the buffer control unit 2 and the input processing unit 3
Figure (b) According to the cylinder control table display (S, :4S2:1) shown in the top column, four input data ladders 'l-"2-"3", °゛4°' are sent from cylinder Sl. The input data "°9°" are sequentially read out from cylinder S2 and stored in input buffers A, B, C, D, and H, respectively (FIG. 5(a)). At this point, the magnetic heads are in cylinders 5a-h, and the contents of the buffer control table are shown in the top column (FIG. 5(C)). Next, the comparison processing unit 4 reads the five input data ゛°1°゛～”
9'' are compared using each classification/merging key according to the buffer control table, and the input data in buffer A with the smallest number ゛°1''° is selected as the winner, and the buffer control unit 2
writes this input data “1” to position a on cylinder S2 where the magnetic head is currently located (after inputting the input data ladder °9°), and also outputs it to the empty buffer A. , the next input data from 1nd S2
10” is loaded.

At the same time, the buffer control unit 2 changes the contents of the buffer control table as shown in the next column, and the cylinder control unit 1 accordingly updates the display of the cylinder control table to 3:2 as shown in the next column. do. Next, a similar comparison operation is performed, and the input data with the lowest number ``2'' is determined to be the winner, and is taken out from buffer B and written into cylinder S, after the manual data ``lO'° on top is vacant. Then, the next input data "'II" on cylinder S2 is read into buffer B, and the displays of the buffer control table and cylinder control table are updated as shown in the next columns.Furthermore, similar operations are performed. When the magnetic head is located at the position on the cylinder S2, the input data on the cylinder S2 has changed to °°bi, °゛2'', 3-12, and the contents of the input buffers A-H. are '10'' each
, "II", "+2°°,"' 4 "" 9°
°, and the cylinder control table display is (S,:1°Sz:4.). Therefore, the cylinder control section 1
refers to the cylinder control table, confirms that the number of input data read from cylinders S1 and S2 is reversed (1; 4), and changes the position of the magnetic head to cylinder S.
Move from position C on cylinder S1 to position C on cylinder S1, then move to cylinder S, and repeat the same process as described above until the number of input data on the cylinder control table is reversed. As shown in the bottom column of the second stage blocks in a) and (b), the contents of the input buffers A to E have the number ratio of input data reversed again. The magnetic head is then returned from position d on cylinder S1 to position on cylinder S2. In this way, the magnetic head is moved only when the number ratio of input data for both cylinders S1 and Sz on the cylinder control table is reversed, and reading and writing are performed by sliding sequentially on the same cylinder as much as possible. As shown in FIG. 6, only four movements are required to complete the merging, and the seek time can be greatly reduced.

〔Effect of the invention〕

As explained above, the present invention initially reads n-1 pieces of data from one cylinder and one piece of data from the other cylinder into n buffers, and stores these n pieces in key order. The data of the cylinder where the magnetic head is located at that time is fetched and the comparison is repeated. When the data of the other cylinder is only 1 out of n, the magnetic head By moving the head to the other cylinder and performing the same process again, the number of times the magnetic head moves between cylinders can be reduced as much as possible, saving seek time and improving the efficiency of sorting and merging processing. It has the effect of being able to.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the disk data classification/merging processing method of the present invention, and FIGS. 2(a) and (b)
) are configuration diagrams of the cylinder control table and buffer control table used in this embodiment, respectively. FIG. 3 is a flowchart showing the processing procedure according to this embodiment. FIG. Figures 5(a), 5(b), and 5(c) show examples of initial read data and written data after merging processing, respectively, for input buffers A, B, C, and D in the case of Figure 4. ,H
FIG. 6 is a diagram showing the movement of the magnetic head between cylinders S1 and 82 in the case of FIG. 4, and FIG. FIG. 3 is a diagram illustrating an example of read processing using a disk data classification/merging processing method. DESCRIPTION OF SYMBOLS 1... Cylinder control part, 2... Buffer control part, 3... Input processing part, 4... Comparison processing part, 5
...Output processing section, 6.7...Cylinder system (wholesale table column, 8.9...
- Buffer control table columns, 31 to 37... Steps, A, B, C, D, E... Input buffer, Sl, S2.
...Cylinder, a, b, c, d...position. Patent applicant: NEC Office Systems Co., Ltd. Agent: Susumu Uchihara, patent attorney Figure 5 (b) Figure (C) Figure 5

Claims (1)

[Scope of Claims] 1. Disk data classification and merging of data arranged in advance on two different cylinders of a disk device according to keys for classification and merging that each has. In this merging method, first, n-1 pieces of data are read from one cylinder, and 1 piece of data is read from the other cylinder, according to the display of the cylinder control table that indicates the number of data to be read initially from two cylinders. Read the n data into each of the n input buffers. At the same time, write the input buffer address and classification/merging key for each of the n data read into the buffer control table, and write the n data read into the input buffer. data are compared using the keys of the buffer control table, and as a result, priority data is determined according to a certain key order, and inputted from the input buffer into the previously outputted space on the cylinder side where the magnetic head is currently located; Next, the next order of data is read from the cylinder that accepted the priority data into the input buffer that has become empty after the priority data has been output, and the number of data in the cylinder control table and the buffer control table are adjusted according to the processing content. Update the contents, and then check the cylinder control table and repeat the same process until the display changes from the initial data count ratio of the two cylinders, n-1:1, to 1:n-1. , Next, the magnetic head is switched from the cylinder where the magnetic head is currently located to the other cylinder, and the same data input/output processing and magnetic head switching are performed again to complete the classification and merging of data on the cylinder. Disk data classification/merging processing method that repeats until