JPH01144136A - Data access system - Google Patents

Abstract

PURPOSE:To realize a highly effective data access system by performing the accesses to plural attributes in two times, i.e., the continuous and discrete accesses regardless of the number of attributes. CONSTITUTION:When data are stored, the data given from outside are sent to a data value memory part 11. An attribute value separating part 12 divides the attribute value of the data stored in the part 11 into the continuous and discrete attributes. The separated continuous attribute value is stored into a continuous multi-dimensional data table 14 via a continuous attribute value storing part 13. Then the continuous attribute value is stored in a discrete multi-dimensional data table 16 via a discrete attribute value storing part 15 based on the control value used by the part 13 for storage and the discrete attribute value. Thus the data accesses using plural attributes are possible with high efficiency.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data access method, and particularly to a data access method in a computer system having files.

[Conventional technology]

Conventionally, in a data access method using a file, data is stored in a file in ascending or descending order of predetermined key attribute values.

To access attributes other than key attributes when the data contains not only key attributes but also attribute values other than key attribute values, access the data sequentially from the beginning until a value of the non-key attribute that satisfies the condition appears. Must read.

To solve this problem, there is a method of having an index for attributes other than key attributes and accessing data from that index.

[Problem that the invention seeks to solve]

However, in the above method, since an attribute index file exists for each attribute, data access using multiple attributes increases processing time and has drawbacks such as an increase in the overall file size. be.

An object of the present invention is to solve the problems of such conventional methods and to provide an access method that efficiently accesses data using a plurality of attributes. [Means for solving the problems] The data access method of the present invention is a data access method for an information processing device that accesses a file holding data consisting of one or more attribute values. means for dividing the continuous attribute values into a plurality of continuous attribute values and a plurality of discrete attribute values; means for storing the continuous attribute values in a multidimensional data table; It has a means for saving it to a table.

[Effect]

According to the present invention, attribute values constituting data are classified into continuous type and discrete type, and continuous type attributes are first stored in a multidimensional file, and then discrete type attributes are stored in a multidimensional file. . For this reason, data can be obtained directly for continuous attributes that are accessed, and for discrete attributes that are accessed with specific values.
For attributes, data can be accessed directly after the continuous attribute has been accessed.

〔Example〕

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

FIG. 1 is a block diagram showing a data access method according to an embodiment of the present invention. In FIG. 1, one embodiment of the present invention first transfers data from the outside to a data value storage unit 11 when storing data. The attribute value separation unit 12 divides the attribute values in the data in the data value storage unit 11 into continuous attributes and discrete attributes. The divided continuous attribute values are stored in the continuous multidimensional data table 14 in the continuous attribute value storage section 13. An example of a multidimensional data table is Grid Phi/I/ (Niebergel t, J et al.
he Grid File: An Adaptable
Symmetric Multi-Key Fill
eStructure, ACM 'I'rans,
on Database 8systems9,x,
pp, 4c+a-so6(t9s4)) can be used. For example, in order to save the continuous attribute values A1 and A2 of the eight pieces of data shown in FIG. 2, the grid file shown in FIG. 3 may be used.

This grid file specifies each grid by a value indicating the boundary of attribute values for each of attributes A1 and A2,
Each grid stores management values for storing discrete attributes. Of course, continuous attribute values can be stored in a continuous multidimensional data table, but since a separate data storage file is used, only management values for discrete attributes can be saved as in this example. That's fine.

Using the management values used when storing in the continuous attribute value storage section 13 and the divided discrete attribute values, the discrete attribute value storage section 15 stores them in the discrete multidimensional data table 16. As an example of the multidimensional data table used here, any grid file can be used. Figure 2 (For the data example shown, Figure 4 shows an example of how it is stored in a discrete multidimensional data table. Figure 4 shows the storage status of the first three data shown in Figure 2. It shows.

As shown in FIG. 4, the discrete multidimensional zone table is managed for each grid of the continuous data table, and the storage location is determined using the management value used when saving the continuous attribute. An example of a method for determining the storage location is to directly set the management value as the address from the beginning of the discrete data table. Another example is a method in which the hashing value of the management value is the address from the beginning of the discrete data table. Each grid of the discrete data table stores an address in a file that stores data corresponding to this table. This address is selected from addresses indicating free areas within the file.

The data storage section 17 stores the data in the disk device 18 at the obtained address.

Next, we will explain how to extract data. The data retrieval conditions are transferred to the condition storage area 21.

The condition check unit 22 divides the conditions into conditions related to continuous attributes and discrete attributes. When the conditions include conditions related to continuous attributes, all conditions related to continuous attributes are passed to the continuous attribute access unit 23, and management values according to the specified conditions are extracted from the continuous multidimensional data table. To use a multidimensional data table, for example, when a grid file is used in the example shown in FIG. 2, the boundaries of AI and A2 can be compared with the continuous attribute values of the data to find a grid that includes these attributes. Once the grid is determined, the control values for that grid can be easily obtained. When the extraction conditions shown in Fig. 5 are used, digrids 2 and 5 are found based on the condition of A2, and the control value is 2 (=4) and the control value is 5 (=14).
is obtained.

When the retrieval condition does not include a continuous type attribute or when the continuous type attribute access is completed, the address of K is retrieved by the discrete type access unit 24 using the condition related to the discrete type. The discrete type multidimensional data table 16 can be accessed by K in the same way as in the case of the continuous type. Under the conditions shown in FIG. 5, the management value kl obtained by the continuous access unit 23 and 2 are used in the discrete access unit 24. First, regarding kl, kl
The discrete multidimensional data table starting from the address indicated by is accessed using the discrete attribute condition, and 1, 1 is found in the grid. Address G stored in that grid
l (=2) is one of the areas that accumulates the required data. Next, for %2, the grid and the address G2 (=6) within it are found. The area at the address indicated by G2 contains data that satisfies the condition. The data retrieval unit 25 retrieves areas 01 and 02, and data satisfying the conditions within these areas is transferred to the outside.

In the explanation so far, we have not defined discrete attributes or continuous attributes, but this difference is due to the difference in access to attributes, and the attributes that are accessed frequently are defined as continuous attributes. Attributes whose complete attribute values are frequently specified are called discrete attributes.

〔Effect of the invention〕

As explained above, the data access method of the present invention is
Compared to conventional data access methods, efficient access is possible because access to multiple attributes requires only two accesses: continuous access and discrete access, regardless of the number of attributes. Furthermore, both continuous and one-discrete multidimensional data tables have the advantage of having a small capacity that can be stored in the primary memory of a computer.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 is a diagram showing an example of data used in this embodiment, and Fig. 3 is a continuous type of data to be stored used in this embodiment. Figure 4 is a diagram showing storage in a data table, Figure 4 is a diagram showing storage in a discrete data table of data to be accumulated used in this example, Figure 5 is an example of retrieval conditions used in this example FIG. 11... Data value storage unit, 12... Attribute value separation unit, 13... Continuous attribute value storage unit, 1
4... Continuous multidimensional data table, 15...
・Discrete attribute value storage unit. 16... Discrete multidimensional data table, 17...
...Data storage unit, 18...Disk device, 2
1... Condition storage area, 22... Condition check section, 23... Continuous attribute access section, 24
. . . Discrete type access section, 25 . . . Data retrieval section. Agent Patent Attorney Susumu Uchihara Diagram 2 A2 Guzotsu V771
$3M

Claims (1)

[Claims] In a data access method for an information processing device that accesses a file that holds data consisting of one or more attribute values, when data is stored, individual data to be stored is combined with a plurality of continuous attribute values and a plurality of continuous attribute values. means for dividing the continuous attribute values into discrete attribute values; means for storing the continuous attribute values in a multidimensional data table; and means for storing the continuous attribute values in the multidimensional data table and then further storing the discrete attribute values in the multidimensional data table. A data access method that enables efficient data retrieval