JPS62290925A - Connection processing system - Google Patents

Abstract

PURPOSE:To ensure the fast connection processing by using the connection information of fixed length. CONSTITUTION:The two-split searching is applied to a tuple to be connected to the head tuple T1 of a table 2 over a connection information table where the connection information on a table 1 are sorted in the order of connection columns. Then an edge tuple S1 of a tuple group of the table 1 having the connection column value coincident with the connection column value 6 of the tuple T1 is detected. The connection is secured between both tuples T1 and S1 with reference to the tuple head pointer of the connection information. When this connection processing is over, the connection information table receives successively accesses as S5-S7 to proceed the connection processing. Then the connection processing is through for the tuple T1 since no coincidence is obtained any more with the connection column value of the T1 in the step S9. Then the connection processing proceeds to a tuple T2. This operation is applied to all tuples of the table 2.

Description

[Detailed Description of the Invention] 3. Detailed Description of the Invention [Field of Industrial Application] The present invention relates to join processing in a database, and is particularly suitable for join processing between tables with variable tuple lengths and column lengths. Regarding the join processing method.

[Conventional technology]

A conventional method is, for example, VLDB '83. Although a basic join processing method is described as described on page 455, no consideration was given to join processing between tables with variable tuple lengths and column lengths.

[Problem that the invention seeks to solve]

An object of the present invention is to provide a method for quickly performing join processing between tables whose tuple lengths and column lengths are variable.

[Means for solving the problem] [In the present invention, from the tuple of table 1 whose tuple length or column length is variable, fixed By creating join information of length and introducing a join information table that sorts it in order of join column values, it is possible to quickly execute join processing between tables with variable tuple lengths and column lengths. .

[Effect]

For example, a binary search is used to quickly find a target tuple in one table and a tuple in another table to be joined, but for this purpose it is necessary to sort the tuples in the tables to be joined. At this time, sorting can be performed faster by sorting relatively short combined information than by sorting relatively long tuples. Also,
Since the tuple length is fixed, it is easy to perform a high-speed binary search on the table of join information.

[Embodiments of the invention]

The present invention will be explained below with reference to examples.

The join processing between tables in a database is to join tuples together when a join condition that there is a relationship between the join column values of the tuples of table 1 and the join column values of the tuples of table 2 is satisfied.

FIG. 1 is a flowchart showing an embodiment of the processing procedure of the present invention. The operation will be explained based on this.

The table with the smaller number of tuples is set as table 1, and join information consisting of a tuple ID, a join column value or a join column value pointer, and a tuple start pointer is created for each tuple in table 1 (step 100), as shown in FIG. 3
The figure shows the structure of the connection information. Figure 2 shows the case where the join information is composed of tuples, L44 combined column values, and tuple head pointers, and Figure 3 shows the case where the join information consists of tuple IDs, join column value pointers, and tuple head pointers. . In the case of FIG. 2, when the combined column value has a variable length, it is set to a fixed length by matching the length of the longest column. Since it is difficult for hardware to handle variable length data, it is necessary to convert variable length data to a fixed length. FIG. 3 introduces a binding column value pointer to have a fixed length. In this way, fixed length combined information is created.

Next, this join information is sorted in order of join column values (101). Sorting relatively short join information takes less time than sorting relatively long tuples of Table 1 in order of join column values. .

Next, pick up one tuple from table 2 (102),
A tuple in table 1 to be combined with that tuple is searched for by binary search (103). At this time, if there are multiple tuples of table 1 to be joined with the target tuple of table 2,
Let's find the most extreme tuple in the group of tuples. Furthermore, in general, when performing a binary search, it is faster to perform a binary search on fixed-length tuples than to perform a binary search on variable-length tuples. In the present invention, this can be achieved by introducing a fixed-length join information table and performing a binary search on it.

Once the tuples to be joined are found, the target tuples in table 2 and the tuples in table 1 are joined. At this time,
By using the tuple head pointer of the join information, the tuples to be joined in table 1 can be accessed at high speed, so that high-speed tuple-to-tuple joining processing becomes possible (104).

Next, sequentially refer to the join information on the join information table until it is no longer a join tuple (step 104).
The process is repeated (105).

In this way, when the join process for one target tuple in table 2 is completed, the above procedure is repeated for the next target tuple, and the process ends when the process is completed for all tuples in table 2 (106 and 107).

FIG. 4 is an explanatory diagram of a combination process when using the combination information of FIG. 2. In this example, the join information table in which the join information of table 1 is sorted into join columns is searched for a tuple to be joined with the first tuple (T1) of table 2. Then, the combined column value of T1 (6
) is searched for the endmost tuple (Sl) of the group of tuples in table 1 that has a join column value that matches. Then, with reference to the tuple head pointer of the combination information, the tuple T1 and tuple S1 are combined. After this is completed, create the join information table in 35. S6. S7 is sequentially accessed to perform the join process, and in S9, since the value does not match the join column value of TI, the join process of T1 is ended, and then the process moves to the join process of T2. This is executed for all tuples in table 2.

FIG. 5 is an explanatory diagram of the combining process when using the combining information of FIG. 3, and the operation is almost the same as that shown in FIG. 4.

As described above, by introducing the join information table as shown in FIGS. 2 and 3, high-speed join processing becomes possible.

At this time, the information of table 1 and the joint information table of table 1 may exist in separate buffers such as the main memory and the buffer in the controller 1-roller.

〔Effect of the invention〕

According to the present invention, high-speed combining processing is possible by introducing fixed-length combining information. Binary search is used to quickly find the tuples in Table 1 that should be combined with the target tuples in Table 2.

To do this, it is necessary to sort the tuples in table 1. At this time, sorting can be performed faster by sorting relatively short combined information than by sorting relatively long tuples. For example, set the tuple length to 20
0 (B), and the combined information length is 10 (B), then 20
Sorting can be done twice as fast. Furthermore, when performing a binary search, if the tuple length is variable, it is difficult to perform the binary search at high speed. On the other hand, it is easy to perform a binary search on a table of fixed-length binding information. Furthermore, by adding the tuple head pointer as the combination information, the actual process of combining tuples can be performed at high speed.

[Brief explanation of drawings]

FIG. 1 is a flowchart showing one embodiment of the present invention, and FIG.
3 and 3 are explanatory diagrams of combined information, and FIGS. 4 and 5 are explanatory diagrams of combined processing. 200...Join information (when using a join column value). 201...Join information (when using a join column value pointer).

Claims (1)

[Claims] 1. When performing table-to-table join processing, fixed-length join information including information that allows easy access to join column values and information that allows easy access to tuples is used. join processing method. 2. The join according to item 1, characterized in that instead of information that allows easy access to the join column value, fixed-length join information is used using a join column value that is a fixed-length join column value from a variable-length join column value. Processing method.