WO2016117739A1

WO2016117739A1 - In-memory database-based data management system and method

Info

Publication number: WO2016117739A1
Application number: PCT/KR2015/000831
Authority: WO
Inventors: 정성훈; 이호진; 한혁; 진성일
Original assignee: ㈜리얼타임테크
Priority date: 2015-01-23
Filing date: 2015-01-27
Publication date: 2016-07-28
Also published as: KR101640733B1

Abstract

The present invention relates to an in-memory database-based data management system comprising: a data input module for generating a node table, by inputting in a node record node data and link record identification information of a link connected to the node data, and for generating a link table by inputting in a link record link data and node record identification information of a node connected to the link data; a data storage module for constructing a network table that comprises the node table or link table; a pointer calculation module for extracting link record identification information and node record identification information from the node table and link table, respectively, and changing into a memory address value the node record identification information and link record identification information, respectively, thereby setting same as a start pointer of a corresponding record; and a control module for performing record access control operation so as to perform a record search by using the node record identification information or record identification information, and for performing information management operation such that the node record identification information or record identification information does not change. Therefore, the present invention enables adding of record identification information, which is a logical record address value that is to be mutually referred between a node table and a link table, to a separate column and then storing same in each table, thereby enabling a path search in a short time while confirming connectivity information between a node and a link by means of a direct record access using record identification information during a network path search. And the present invention enables memory pointer calculation and thereby secures extremely fast response speed by having record searching time through an existing index completely removed.

Description

In-memory database based data management system and method

The present invention relates to an in-memory database-based data management system and method thereof, and more particularly, to an in-memory database-based data management system capable of performing a high-performance network path search through direct record access in a database management system. And to a method thereof.

DBMS (Database Management System), the basic software for organizing and archiving data for effective use, is a software package that combines functions such as adding, changing, deleting, and retrieving data, which are necessary steps for managing a database.

In-Memory Computing is when an application loads data for operation into main memory rather than a hard disk to perform a service. In-memory computing uses a memory area that was used only as an area for computation as a space for storing and processing a large amount of data.

These in-memory computing technologies can be classified into three categories: in-memory application platform, in-memory data management, and in-memory messaging. In particular, an in-memory database is a database management system that is installed and operated in the main memory of a data store, and data access and processing speed are faster than those installed on a disk.

In in-memory data management, the in-memory database management system stores the records in main memory and transfers the data directly from the main memory to the CPU. Quickly search through the index

When implementing a network path search algorithm in an in-memory database, a large number of node data and link data must be accessed and read for path search. At this time, the in-memory database management system must perform a path search algorithm by searching the records of the node table and the link table through index search each time for mutual data access between the node table and the link table.

As prior art data, Korean Patent No. 10-1260165 discloses a technical content of an open database management system.

In a conventional open database management system, a record is composed of a key, and a column is created by designating a plurality of regions each consisting of a set of values as a table, which is implemented as a database management system of key-value pairs, and is composed of a domain key and a record key. You can also link between records in tables that exist in different accounts or regions by referring to the link table composed of pairs of IDs.

Here, a key is a field or fields in a table that are indexed so that you can quickly find the data you want in the database.

As a result, the prior art performs record retrieval by using an index structure using a key for linking between records of tables existing in different accounts or regions, and thus it takes a lot of time for the record retrieval. When the path search performance is inevitably deteriorated, there is a problem that it takes a long time to obtain the final result.

According to the present invention, the record identification information, which is a logical address value of a record that should be cross-referenced between node data and link data, is stored separately in a network table so that the record can be directly accessed through the record identification information when searching for records between nodes and links. It provides a memory database-based data management system and method thereof.

In embodiments, an in-memory database based data management system may generate a node table by inputting node data and link record identification information of a link connected to the node data, and generating the link data and the link in a link record. A data input module for inputting node record identification information of a node connected with data to generate a link table; A data storage module for constructing a network table including the node table or a link table; A pointer operation for extracting link record identification information and node record identification information from the node table and the link table, respectively, and changing the node record identification information and the link record identification information to address values of a memory and setting them as starting pointers of the corresponding records. module; And a control module for performing a record access control operation to perform a record search using the node record identification information or the record identification information, and performing an information management operation so that the node record identification information or the record identification information is not changed. It features.

The data management system based on the in-memory database may further include a path search module for accessing a corresponding record directly using the start pointer to check connectivity information between a node and a link and performing a path search operation.

The data management system based on an in-memory database may further include a memory compression module configured to perform a compression operation to prevent fragmentation of the memory while maintaining data consistency of the data storage module. The memory compression module controls the memory compression module to perform a compression operation on the remaining tables except the network table through an information management operation.

Meanwhile, the node table adds a link identification column for inserting link record identification information of link data connected to each node, and the link table inserts node identification column for inserting node record identification information of a start node and an end node of each link. It is characterized by adding the.

Among the embodiments, an in-memory database-based data management method includes an in-memory database-based data management method for constructing network data by an in-memory database management system (DBMS). The node table is generated by inputting node data and link record identification information of the link connected to the node in the node record, and the link table is generated by entering link data and node record identification information of the node associated with the link in the link record. Creating a table; A data storage step of storing a network table including the node table or a link table on a memory; A pointer operation for extracting link record identification information and node record identification information from the node table and the link table, respectively, and changing the extracted node record identification information and link record identification information to memory address values and setting them as starting pointers of the corresponding records. step; And a record access step of confirming the connectivity information between the node and the link using the node record identification information or the record identification information and performing a record access control operation to perform a record search.

The table generating step may include a link identification column for inserting link record identification information of a link connected to each node in the node table, and inserting node record identification information of a start node and an end node of each link in the link table. The node identification column may be added.

The table generating step may include inputting node data into the node table and setting the link identification column as an initialization value; If a start node and an end node for a link of the link data exist in the node table, the corresponding link data is input to the link table, and node record identification information of the start node and the end node of the link is entered in the node identification column. Inserting each; And inserting link record identification information connected to the start node in the link identification column of the node record for the start node in the node table, and identifying the link record connected to the end node in the link identification column of the node record for the end node. And inserting the information.

Meanwhile, in the in-memory database-based data management method, when searching for a network path, a node record including node data for initial access is searched in the network table, and the link record connected to the node is identified in the searched node record. And extracting information, and using the extracted link record identification information as a start pointer, directly accessing the corresponding record to perform a path search operation.

In addition, the in-memory database-based data management method may perform a compression operation to prevent fragmentation of memory for the tables other than the network table so that the node record identification information and the link record identification information are not changed. It further comprises a memory compression step to perform.

In the pointer operation step, a pointer operation is performed on the link record identification information stored in the node record to be changed to the start pointer of the corresponding link record, and the node record identification information stored on the link record is converted to the start pointer of the corresponding node record. And cross-reference between the node data and the link data.

In the in-memory database-based data management system and method of the present invention, the record identification information by adding the record identification information, which is the logical address value of the record that should be cross-referenced between the node table and the link table, in a separate column and storing it in each table Through the direct access to the corresponding record, in particular, it is possible to perform the path search in a short time while checking the connectivity information between the node and the link through the direct record access using the record identification information when searching the network path.

Since a method of accessing a record by using the record identification information of the present invention is a memory pointer type operation operation, the record retrieval time through the index is completely removed, thereby ensuring a very fast response speed.

1 is a diagram illustrating a configuration of an in-memory database based data management system according to an exemplary embodiment of the present invention.

2 is a flowchart illustrating an in-memory database based data management method according to an embodiment of the present invention.

FIG. 3 is a diagram illustrating a record direct access method of FIG. 2.

4 is a flowchart illustrating a table generation process of FIG. 2.

Description of the present invention is only an embodiment for structural or functional description, the scope of the present invention should not be construed as limited by the embodiments described in the text. That is, since the embodiments may be variously modified and may have various forms, the scope of the present invention should be understood to include equivalents capable of realizing the technical idea. In addition, the objects or effects presented in the present invention does not mean that a specific embodiment should include all or only such effects, the scope of the present invention should not be understood as being limited thereby.

On the other hand, the meaning of the terms described in the present invention will be understood as follows.

Terms such as "first" and "second" are intended to distinguish one component from another component, and the scope of rights should not be limited by these terms. For example, the first component may be named a second component, and similarly, the second component may also be named a first component.

When a component is referred to as being "connected" to another component, it should be understood that there may be other components in between, although it may be directly connected to the other component. On the other hand, when a component is referred to as being "directly connected" to another component, it should be understood that there is no other component in between. On the other hand, other expressions describing the relationship between the components, such as "between" and "immediately between" or "neighboring to" and "directly neighboring to", should be interpreted as well.

Singular expressions should be understood to include plural expressions unless the context clearly indicates otherwise, and terms such as "comprise" or "have" refer to features, numbers, steps, operations, components, parts, or parts thereof described. It is to be understood that the combination is intended to be present and does not exclude in advance the possibility of the presence or addition of one or more other features or numbers, steps, operations, components, parts or combinations thereof.

In each step, an identification code (e.g., a, b, c, etc.) is used for convenience of description, and the identification code does not describe the order of the steps, and each step clearly indicates a specific order in context. Unless stated otherwise, they may occur out of the order noted. That is, each step may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the reverse order.

All terms used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. Generally, the terms defined in the dictionary used are to be interpreted as being consistent with the meanings in the context of the related art, and should not be interpreted as having ideal or excessively formal meanings unless clearly defined in the present invention.

Referring to FIG. 1, an in-memory database based data management system includes a data input module 110, a data storage module 120, a pointer operation module 130, a control module 140, a path search module 150, and Memory compression module 160.

The data input module 110 inserts node data and link record identification information LINK_RID of a link connected with a node of the node data to each node record to generate a node table. In addition, the data input module 110 inserts link data and NODE_RID of each link connected node into each link record to generate a link table. At this time, since the node table and the link table have their own RID information, the node table and the link table inform their own RID values according to requests of tables to be referred to each other.

The node table and the link table are tables for organizing network data. As shown in Table 1 and Table 2, a node table and a link table are sets of records composed of columns. At this time, the node data is represented through a node list in which each node is represented as a node record. Node records contain one or more fields that are unique identifiers for that node, but may include other attribute fields that provide information about the node. Link data is represented through a link list (or link matrix) that identifies at least two nodes (start node and end node) but can also display information such as link strength, link direction, and link type. .

Table 1

TABLE 2

In Table 1 and Table 2, the node data is NODE_ID as identification information for identifying the node record by the user and has a value such as coordinates, and the link data is LINK_ID as identification information for identifying the link record by the user. It has values such as start node ID, end node ID, and cost. RID, on the other hand, is the logical address translation of memory from the system.

The data storage module 120 stores a network table including a node table and a link table generated by the data input module 110 in a main memory that is a data storage of a DBMS. In addition to the network table, the main memory stores tables related to various data.

The pointer operation module 130 extracts the LINK_RID from the node table stored in the data storage module 120, extracts the NODE_RID from the link table, and changes the extracted NODE_RID and LINK_RID to logical address values of the memory, respectively, to start the corresponding record. Set by pointer. In other words, if the pointer operation is performed on the LINK_RID information stored in the node record, the start pointer of the link record can be obtained. If the pointer operation is performed on the NODE_RID information stored in the link record, the start pointer of the node record can be obtained. Cross reference is possible.

The control module 140 performs a record access control operation to directly access a record using NODE_RID and LINK_RID when searching for a network path, and performs an information management operation so that the NODE_RID and LINK_RID of the network table are not changed.

The path search module 150 may return a record through the pointer calculation module 130 and request a record to be searched by the pointer calculation module 130. That is, the path search module 150 directly accesses the corresponding record using the start pointer, checks the connectivity information between the node and the link, and proceeds with the path search operation. In this case, the path search module 150 may use a path search algorithm such as a Dijkstra algorithm or an A * algorithm.

The memory compression module 160 performs a compression operation for each page to prevent fragmentation of the memory while maintaining data consistency of the data storage module 120.

In this case, when the NODE_RID and LINK_RID of the network table are changed during the compression operation of the memory compression module 160, the control module 140 changes the location on the physical memory. Compress only the rest of the tables.

2 is a flowchart illustrating an in-memory database based data management method according to an embodiment of the present invention, and FIG. 3 is a diagram illustrating a record direct access method of FIG. 2.

Referring to FIG. 2, in the in-memory database-based data management method, a node table is generated by inputting node data and LINK_RID of a link connected to a node in each node record through the data input module 110. The link table is generated by inputting link data and NODE_RID of a node connected to the link data in each link record (S10).

As such, the node table has a link identification column for inserting the LINK_RID of the link connected to each node, and the link table has a node identification column for inserting START_NODE_RID and END_NODE_RID, which are node record identification information of the start and end nodes of each link. Is added.

The network table including the node table and the link table thus generated is stored in the main memory (S20).

The pointer operation module 130 extracts the LINK_RID and the NODE_RID from the node table and the link table, respectively, and changes them into memory address values and sets them as the start pointers of the corresponding records. (S30) If NODE_RID and LINK_RID are set as the start pointers of the memory, In addition, the memory pointer type of arithmetic operations allows direct access to records, which ensures faster response times than record searches using existing index structures.

That is, in the node table, the LINK_RID of the link connected to each node is a start pointer, and the link table is the start pointer of NODE_RID of the start node and the end node of the end node connected to each link. Thus, by retrieving a particular node ID, you can extract the LINK_RID while identifying the link associated with this node, and access the link record with this LINK_RID as the starting pointer directly.

Therefore, the path search module 150 may directly access the corresponding record using the start pointer to check the connectivity information between the node and the link and perform the shortest path search operation at high speed.

Referring to Tables 1 and 2 above, the node table inputs a node ID and a value of node data, and a link identification column for inputting a LINK_RID connected to the node is added. In the link table, a node ID column for inputting a link ID and a value of link data and inputting NODE_RIDs of a start node and an end node of the link is added.

As shown in FIG. 3, the path search module 150 searches for a record to be visited by searching for a node ID or a link ID when the first record is accessed, and then starts a path search algorithm.

The path search module 150 extracts the NODE_RID and LINK_RID for the linked node and link from the record retrieved through the pointer operation module 130, and directly accesses the record using the extracted NODE_RID and LINK_RID as the starting pointer of the corresponding record. Proceed with the path search operation.

For example, a route search retrieves a node with ID 104 to retrieve the first access node, retrieves a value of '004' from the LINK_RID value of the retrieved node record to visit a link that is connected to node 104, You can directly access the link record by converting it to the start pointer of the link record through pointer operation.

The main memory, which is the data storage module 120, divides the entire memory space into pages of the same size and stores a plurality of data or tables. However, since data or tables stored in memory are stored in different sizes, page fragmentation, which is an empty memory space, occurs in a page.

Therefore, the memory compression module 160 performs a compression operation to make a memory space that can be used by gathering fragmented spaces scattered here and there in one place. At this time, the control module 140 performs an information management operation so that the network table does not perform a compression operation so that NODE_RID and LINK_RID of the node table and the link table are not changed. Therefore, the memory compression module 160 performs the compression operation only on the data or the table except the network table by the control module 140.

4 is a flowchart illustrating a table generation process of FIG. 2.

Referring to FIG. 4, NODE_RID and LINK_RID stored in the node table and the link table are address values that should be referred to each other. Therefore, the data input module 110 should store address values for records connected to the node identification column and the link identification column.

To this end, the data input module 110 inputs node data into the node record, and sets the address value of the link identification column to an initialization value (000). (S11, S12)

Check whether the start node and the end node of the specific link (LINK1) exist in the node table (S13), enter the link data in the link record, and then enter the start node and the end node of the link in the node table in the node identification column. Enter START_NODE_RID and END_NODE_RID respectively (S14).

The data input module 110 updates the initialization value set in the link identification column of the record of the start node in the node table to the LINK1_RID of the link connected to the start node, and updates the initialization value set in the link identification column of the record of the end node to the end node. Update to LINK1_RID of the linked link (S15).

In this manner, the data input module 110 inputs data into the node table and the link table, respectively, and stores an RID which is an address value that should be cross-referenced to the node identification column and the link identification column.

As such, an in-memory database-based data management system and method thereof according to an embodiment of the present invention are linked with node record identification information, which is a logical address of an object record to be visited for direct record access to the node table and the link table. A data storage structure that adds record identification information to a column is defined, and a path search algorithm can be performed based on the high performance network path search.

Although described above with reference to a preferred embodiment of the present invention, those skilled in the art will be variously modified and changed within the scope of the invention without departing from the spirit and scope of the invention described in the claims below I can understand that you can.

Claims

A node table is generated by inputting node data and link record identification information of a link connected to the node data in a node record, and a link table is generated by inputting link data and node record identification information of a node connected to the link data in a link record. A data input module;

A data storage module for constructing a network table including the node table or a link table;

A pointer operation for extracting link record identification information and node record identification information from the node table and the link table, respectively, and changing the node record identification information and the link record identification information to address values of a memory and setting them as starting pointers of the corresponding records. module; And

And a control module configured to perform a record access control operation to perform a record search using the node record identification information or the record identification information, and perform an information management operation so that the node record identification information or the record identification information is not changed. In-memory database based data management system.
The method of claim 1,

And a path search module for directly accessing the corresponding record using the start pointer to check connectivity information between the node and the link and performing a path search operation.
The method of claim 1,

Further comprising a memory compression module for performing a compression operation to prevent fragmentation of the memory while maintaining the data consistency of the data storage module,

And the control module controls the memory compression module to perform a compression operation on a table other than the network table through an information management operation.
The method of claim 1,

The node table adds a link identification column for inserting link record identification information of link data connected to each node,

And the link table adds a node identification column to insert node record identification information of a start node and an end node of each link.
An in-memory database based data management method for building network data by an in-memory database management system (DBMS),

A node table is created by entering node data and link record identification information of the link associated with the node in the node record, and a link table is generated by entering link data and node record identification information of the node associated with the link in the link record. Generation step;

A data storage step of storing a network table including the node table or a link table on a memory;

A pointer operation for extracting link record identification information and node record identification information from the node table and the link table, respectively, and changing the extracted node record identification information and link record identification information to memory address values and setting them as starting pointers of the corresponding records. step; And

In-memory database-based data management, characterized in that it comprises a record access step of performing a record access control operation to check the connectivity information between the node and the link using the node record identification information or record identification information and perform a record search. Way.
The method of claim 5,

The table generation step,

Add a link identification column to insert link record identification information of a link connected to each node in the node table,

And a node identification column for inserting node record identification information of a start node and an end node of each link to the link table.
The method of claim 6,

The table generation step,

Inputting node data into the node table and setting the link identification column to an initialization value;

If a start node and an end node for a link of the link data exist in the node table, the corresponding link data is input to the link table, and node record identification information of the start node and the end node of the link is entered in the node identification column. Inserting each; And

Inserting link record identification information connected to the start node into the link identification column of the node record for the start node in the node table, and link record identification information connected to the end node in the link identification column of the node record for the end node In-memory database-based data management method comprising the step of inserting.
The method of claim 5,

When searching for a network path, search for a node record including node data for initial access in the network table, extract link record identification information connected to the node from the searched node record, and start the extracted link record identification information. And a path search step of performing a path search operation by directly accessing a corresponding record using a pointer.
The method of claim 5,

And a memory compression step of performing a compression operation to prevent fragmentation of the memory for the remaining tables except the network table so that the node record identification information and the link record identification information are not changed. In-memory database based data management method.
The method of claim 5,

The pointer operation step,

A pointer operation is performed to change the link record identification information stored in the node record to the start pointer of the corresponding link record, and a node operation is performed on the node record identification information stored in the link record to be changed to the start pointer of the corresponding node record to link with the node data. In-memory database-based data management method characterized in that the data is cross-referenced.