KR102633926B1 - Dual Mode Sharding System and Method for Supporting Client Sharding and Proxy Server Sharding - Google Patents

Abstract

A dual mode sharding system and method supporting client sharding and proxy server sharding are provided. According to an embodiment of the present invention, a dual-mode sharding system supporting client sharding and proxy server sharding includes a client that determines one of the sharding methods of client sharding and proxy server sharding, a proxy server node that includes key distribution information, and Includes a plurality of data nodes and a session manager.

Description

Dual Mode Sharding System and Method for Supporting Client Sharding and Proxy Server Sharding}

The present invention relates to a Database Management System. More specifically, it relates to a distributed database management method and structure supporting dynamic dual-mode sharding.

Recently, as the amount of data has exploded, technology for distributing and storing data to process large amounts of data is rapidly developing. Among them, sharding is a technology that partitions data to distribute and store and process large amounts of data in a relational database.

Conventionally, sharding behavior is initially determined in one way or another by the requirements of the application.

In addition, in the case of the conventional server-side sharding system, it is difficult to expect performance improvement as the number of added data nodes increases, and in the case of the client sharding system, there is a disadvantage in that the cost of rebuilding the system is high when the data distribution policy is changed.

To overcome these shortcomings, by providing a dual-mode sharding system that simultaneously supports server and client sharding systems, performance and system optimization can be achieved through efficient data distribution in two directions according to the requirements of the application.

The goal of the present invention is to enable efficient operation of performance and resources by dynamically selecting and operating sharding operations according to application requirements in a system that supports dual-mode sharding.

In order to solve the above-mentioned technical problems, a dual-mode sharding system supporting client sharding and proxy server sharding according to an embodiment of the present invention includes a client that determines one of the sharding methods of client sharding and proxy server sharding; a proxy server node, including key distribution information; a plurality of data nodes; and a session manager, wherein the client receives a SQL query request by a user input and transmits a request for providing key distribution information to the proxy server node, and the proxy server node responds to the request for providing key distribution information. The key distribution information is transmitted to the client, the client determines one of the client sharding and the proxy server sharding based on the key distribution information, and the session manager determines the client sharding. In this case, the client is connected to a single data node among the plurality of data nodes, and the session manager connects the client with two or more data nodes among the plurality of data nodes when proxy server sharding is determined. When performing the connection of the proxy server node, the session manager connects the proxy server node and the two or more data nodes in parallel rather than sequentially. Processed simultaneously, the session manager does not perform connection to the plurality of data nodes in advance, but connects to the data node to be connected after the sharding method is determined.

The proxy server node operates as a proxy server and includes shard configuration information, and the data node is a relational data management system and may perform the function of storing data.

The client is a server running Vado Keeper, and the Vado Keeper receives the SQL query request, transmits a request for providing the key distribution information to the proxy server node, and based on the key distribution information By determining whether a connection to a specific data node is necessary, one of the client sharding and the proxy server sharding can be determined.

The vado keeper, when the client sharding is determined, transmits a connection request to the data node and receives a response to the connection request from the data node, and the proxy server node, when the proxy server sharding is determined, In this case, the bar can receive a connection request from the keeper, check the key value in response to the connection request, and simultaneously request a connection to two or more data nodes among a plurality of data nodes in parallel.

The proxy server node and the data node can be built independently and can be redundant.

The plurality of data nodes generate a respective data node configuration for each of the plurality of data nodes, create a respective shard table for each of the plurality of data nodes, and generate a respective shard table for each of the plurality of data nodes. It can be configured by creating an index for, setting the sharding method for each shard table, and setting the sharding operation for data input and the input method to the data node.

According to another embodiment of the present invention, a dual mode sharding method using a dual mode sharding system supporting client sharding and proxy server sharding, wherein the dual mode sharding system uses any one of client sharding and proxy server sharding. It includes a client making a decision, a proxy server node containing key distribution information, a plurality of data nodes, and a session manager, wherein the client receives a SQL query request by a user input, and the proxy transmitting a request to provide key distribution information to a server node; The proxy server node transmitting the key distribution information to the client in response to a request for provision of the key distribution information; determining, by the client, a sharding method among the client sharding and the proxy server sharding based on the key distribution information; and the session manager, when the client sharding is determined, connects the client with a single data node among the plurality of data nodes, and the session manager performs the connection of the plurality of data nodes when the proxy server sharding is determined. A step of connecting two or more data nodes among data nodes with the proxy server node, and in the step of connecting the two or more data nodes with the proxy server node, the session manager, the proxy server The connection between a node and the two or more data nodes is simultaneously processed in parallel rather than sequentially, and the session manager does not perform connection with the plurality of data nodes in advance, and connects the data node to which the sharding method is determined. Perform a connection with .

The vado keeper, when the client sharding is determined, transmits a connection request to the data node and receives a response to the connection request from the data node, and the proxy server node, when the proxy server sharding is determined, In this case, the bar can receive a connection request from the keeper, check the key value in response to the connection request, and simultaneously request a connection to two or more data nodes among a plurality of data nodes in parallel.

When the client sharding is determined, the vado keeper includes the step of sending a connection request to the data node, receiving a response to the connection request from the data node, and the proxy server node, the proxy When server sharding is determined, receiving a connection request from the Vado Keeper, checking a key value in response to the connection request, and simultaneously requesting a connection to two or more data nodes among the plurality of data nodes in parallel It can be included.

The plurality of data nodes generates a respective data node configuration for each of the plurality of data nodes, creates a respective shard table for each of the plurality of data nodes, and indexes for each shard table. It can be configured by creating a sharding method for each shard table, setting a sharding operation for data input and an input method to a data node.

According to the present invention as described above, the sharding operation is dynamically selected and operated according to the requirements of various applications such as increase in data queries and data integration, enabling efficient operation of performance and resources.

Figure 1 is a diagram showing an embodiment of the structure of a dual mode sharding system.
Figure 2 is a diagram showing an embodiment of operation in a dual mode sharding system.
Figure 3 is a diagram showing an embodiment of an algorithm by which Vado Keeper selects a sharding method.
Figure 4 is a diagram showing an embodiment of the connection relationship between a client, a proxy server node, and a data node.
Figure 5 is a diagram showing an embodiment of parallel processing for the connection between a client and a data node and a connection between a proxy server node and a data node.
Figure 6 is a diagram illustrating an embodiment of a delayed ready connection.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. The present embodiments are merely intended to ensure that the disclosure of the present invention is complete, and that the present invention is not limited to the embodiments disclosed below and is provided by those skilled in the art It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings that can be commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless clearly specifically defined.

The terminology used herein is for describing embodiments and is not intended to limit the invention. As used herein, singular forms also include plural forms, unless specifically stated otherwise in the context. As used in the specification, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other elements in addition to the mentioned elements.

Unless otherwise stated herein, “connected” or “connected” means that one element/feature is directly connected or connected to another element/feature, or indirectly connected or connected through another element/feature. It may include things, and does not necessarily mean that it is directly connected or connected mechanically. Accordingly, while the various schematic diagrams shown in the figures illustrate example arrangements of elements and components, additional intervening elements, devices, features or components may be added (assuming the functionality of the depicted elements is not adversely affected). ) may exist in actual embodiments.

In addition, in this specification, “transmitting” or “receiving” may include not only directly transmitting or receiving information between a sender and a receiver, but also transmitting or receiving information through another object, unless otherwise specified. .

Hereinafter, a dual mode sharding system and a method for supporting the dual mode sharding system according to embodiments of the present invention will be described with reference to the drawings.

Referring to Figures 1 and 3, a dual mode sharding system supporting client sharding (11) and proxy server sharding (21) according to an embodiment of the present invention includes a client (1), a proxy server node (2), and data Includes node (3) and session manager (6).

Referring to Figure 1, the client (1) may be a server on which an application runs, or a server on which Vado Keeper runs, and client sharding (11) and proxy server sharding are performed by Vado Keeper (4). One of the sharding methods in (21) can be determined.

The proxy server node (2) may be a relational database management system (RDMS), may perform the role of a proxy server, may have shard configuration information, and may be the data node (3). It may have data key distribution information.

Data node 3 may be a relational database management system (RDMS) and may perform a data storage role.

Next, with reference to Figure 3, the algorithm by which Vado Keeper selects the sharding method is described.

When the client (1) receives a SQL query request by user input, the bardokeeper (4) of the client (1) can receive the SQL query (user query analysis, S100). At this time, the bardo keeper 4 may transmit a request for provision of key distribution information to the proxy server node 2, and the proxy server node 2 may send key distribution information to the client ( 1) It can be sent to . Afterwards, Vado Keeper (4) can operate in the order of executing SQL_Prepare (S200), checking key column (S300), executing SQL_BindParameter (S400), and checking key value (S500).

At the key value check (S500) stage, the Vado Keeper (4) can determine whether a connection to a specific data node is necessary based on the key distribution information, and to complete the SQL query result, a single data node among a plurality of data nodes can be selected. When connection to a data node is required, client sharding (11) is performed to perform the main client operation, and when connection to two or more data nodes among a plurality of data nodes is required, proxy server sharding (21) is performed. Performs the main proxy server operation. In this case, the sharding operation is dynamically selected and operated as the sharding method that meets the application requirements is selected, enabling efficient operation of performance and resources.

When client sharding (11) is decided, the session manager (6) performs a connection between the client (1) and a single data node among a plurality of data nodes, and the session manager (6) performs proxy server sharding (21). ) If this is determined, connection between two or more data nodes among the plurality of data nodes and the proxy server node can be performed.

Referring to Figure 2, in the main operation of the client, the client's bardo keeper (4) makes a) request (12) to the data node (3), and the data node (3) that receives request (12) a) A) Response (13) can be given to request (12).

In the proxy server main operation (S20), the client's bardo keeper (4) sends a) request (22) to the proxy server node (2), and the proxy server node (2) that receives request (22) sends the Key value By checking, multi-request (24) can be made in parallel operation to two or more data nodes (3) among a plurality of data nodes. c) The data node (3) that received the request (24) can send the execution result to the proxy server node (2) c) response (25). The proxy server node 3 may collect responses received from two or more data nodes 3 among the plurality of data nodes and send a response 23 to the client 1.

Referring to FIG. 4, the dual mode sharding system includes a client, a vado keeper (4), a proxy server node (2), and a plurality of data nodes (3). The client has a shard connection (31) with the proxy server node and the data node, and a meta connection (32) between the proxy server node (2) and the data node (3).

In the present invention, proxy server nodes and data nodes can be built in an independent (Stand Alone) form, and duplication between proxy server nodes and data nodes can be used.

In the present invention, an active database management system (Active DBMS) or a standby database management system (Standby DBMS) can be built on one physical server. Referring to FIG. 4, as an embodiment of the present invention, for example, 8 DBMSs may be configured. In this case, the DBMS that is activated and provides normal service is the activated DBMS, and the DBMS that is provisionally configured is the standby DBMS. Applies to DBMS in use.

As an embodiment of the present invention, using redundancy means configuring a standby DBMS so that when the activated DBMS is damaged, the standby DBMS can be used, and a stable data management technology can be provided through redundancy.

As an embodiment of the present invention, the redundancy operation method is, for example, the redundancy transmitting processor of the activated DBMS reads the Redo Log (Transaction Logging File), processes it into a redundancy log, and sends TCP/RBC to the redundancy receiving processor of the waiting DBMS. It can be transmitted through IP communication. The duplication receiving processor can restore the received duplication log into a transaction log and reflect it in the waiting DBMS. Data changes corresponding to insert/update/delete in the activated DBMS are equally applied to the standby DBMS.

Meanwhile, Stand Alone refers to a case where the standby DBMS is not configured due to lack of cost, etc., and only the activated DBMS can be configured.

Next, as an embodiment of the present invention, the client main operation and proxy server main operation for data input or data inquiry are described.

First, an example of configuring a data node for data input is as follows.

/* Create data node configuration */

create shard node1 with '192.168.0.11',22581;

create shard node1 with '192.168.0.12',22581;

/* Create a shard table for all data nodes */

node [all]

create table tb_test1

(

c1 integer primary key,

c2 char(10);

c3 integer,

c4 char(10);

c5 char(10);

c6 integer,

c7 date );

/* Create index for shard table */

node[all]

create index TB_TEST1_IDX on TB_TEST1 (c3 desc, c1 asc);

/* Set sharding method for table of data node */

alter table tb_test1 shard by AUTOHASH(c1)

(

shard node1,

shard node2

);

/* Sharding operation for data input and input method to data node */

ex) Assuming hash value (node1 = value of 0~4, node2=5~9)

a. insert into tb_test1 values( 0, 'x', 0, 'abc', 'Gananada', 1, sysdate);

b. insert into tb_test1 values( 1, 'x', 0, 'abc', 'Gananada', 1, sysdate);

c. insert into tb_test1 values( 2, 'x', 0, 'abc', 'Gananada', 1, sysdate);

d. insert into tb_test1 values( 3, 'x', 0, 'abc', 'Gananada', 1, sysdate);

e. insert into tb_test1 values( 4, 'x', 0, 'abc', 'Gananada', 1, sysdate);

f. insert into tb_test1 values( 5, 'x', 0, 'abc', 'Gananada', 1, sysdate);

g. insert into tb_test1 values( 6, 'x', 0, 'abc', 'Gananada', 1, sysdate);

h. insert into tb_test1 values( 7, 'x', 0, 'abc', 'Gananada', 1, sysdate);

i. insert into tb_test1 values(8, 'x', 0, 'abc', 'Gananada', 1, sysdate);

j. insert into tb_test1 values( 9, 'x', 0, 'abc', 'Gananada', 1, sysdate);

In the dual mode sharding system of the present invention, as an example, a client main operation for data input may be performed. The main client operation is loading the shard configuration information of the proxy server from Vado Keeper (4) when the application is first run, executing SQL_Prepare on the SQL executed by Vado Keeper (4), checking key columns, and executing SQL_BindParameter(). It includes steps that proceed in order. Vado Keeper (4) can know the data node based on the shard key value of the bind value corresponding to the key, for example, the hash value. Afterwards, a connection is made only to the corresponding data node (Deferred Preparation Connection). You can execute SQL_execute through a connected session and receive a success request under normal circumstances.

As an example in the present invention, the main client operation process for data input is as follows.

/* Operation process for each SQL */

a. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,0) > node1 deferred preparation connection > SQL_execute > SQL_Sucess request

b. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,1) > node1 deferred preparation connection > SQL_execute > SQL_Sucess request

c. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,2) > node1 deferred preparation connection > SQL_execute > SQL_Sucess request

d. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,3) > node1 deferred preparation connection > SQL_execute > SQL_Sucess request

e. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,4) > node1 deferred preparation connection > SQL_execute > SQL_Sucess request

f. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,5) > node2 deferred preparation connection > SQL_execute > SQL_Sucess request

g. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,6) > node2 deferred preparation connection > SQL_execute > SQL_Sucess request

h. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,7) > node2 deferred preparation connection > SQL_execute > SQL_Sucess request

i. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,8) > node2 deferred preparation connection > SQL_execute > SQL_Sucess request

j. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,9) > node2 deferred preparation connection > SQL_execute > SQL_Sucess request

In the dual mode sharding system of the present invention, as an example, proxy server main operation for data input may be performed. An application can establish a connection using a generic driver to a proxy server and execute SQL. The main operation of the proxy server can proceed in order of executing SQL_Prepare, checking key columns, executing SQL_BindParameter(), and selecting nodes.

The main operation of the proxy server is to know the data node based on the shard key value of the bind value corresponding to the key, for example, the hash value. At this time, a native connection (delay ready connection, 100), not a driver connection, is created from the proxy server's session manager to the corresponding data node. You can execute SQL_execute through a connected session and receive a success request in normal cases.

As an example in the present invention, the proxy server main operation process for data input is as follows.

This is the operation process for each SQL.

a. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,0) > node1 native connection > SQL_execute > SQL_Sucess request

b. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,1) > node1 native connection > SQL_execute > SQL_Sucess request

c. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,2) > node1 native connection > SQL_execute > SQL_Sucess request

d. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,3) > node1 native connection > SQL_execute > SQL_Sucess request

e. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,4) > node1 native connection > SQL_execute > SQL_Sucess request

f. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,5) > node2 native connection > SQL_execute > SQL_Sucess request

g. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,6) > node2 native connection > SQL_execute > SQL_Sucess request

h. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,7) > node2 native connection > SQL_execute > SQL_Sucess request

i. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,8) > node2 native connection > SQL_execute > SQL_Sucess request

j. (Query Analyzer)SQL_prepare > Check key column c1 > SQL_bindParameter(1,9) > node2 native connection > SQL_execute > SQL_Sucess request

The following is an example of configuring a shard node for data inquiry.

/* Shard operation and node access method for data inquiry */

ex) Assuming hash value (node1 = value of 0~4, node2=5~9)

a. select * from tb_test1;

b. select * from tb_test1 where c1=0;

c. select * from tb_test1 where c1=4;

d. select * from tb_test1 where c1=5;

e. select * from tb_test1 where c1=9;

In the system of the present invention, the client main operation for data inquiry can be performed. When running the application for the first time, the shard configuration information of the proxy server can be loaded from Vado Keeper. In the case of a client query in a typical invention, query SQL for multiple nodes returns an error or generates a guide alarm on how to use, but the present invention automatically transfers the SQL to the proxy server, so that the SQL is executed in the following procedure without any problem.

Vado Keeper can proceed with executing SQL in the following order: executing SQL_Prepare, checking key column, and executing SQL_BindParameter(). The Vado Keeper can know the data node based on the shard key value of the bind value corresponding to the key, for example, the hash value.

At this time, depending on the bind value of SQL, a connection (lazy ready connection) is created to a single data node, SQL_execute is executed on that data node, and the result is given a response to the client, or SQL is executed for multiple data nodes. To do this, SQL can be transferred to a proxy server, and at this time, the proxy server that has received the SQL transfer can receive SQL results from each data node, collect them, and provide a response to the application.

As an example in the present invention, the operation process and results for each SQL are as follows.

a. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(none) > Shard proxy server (query analyzer) > SQL_bindParameter(none) > node1,2 deferred preparation native connection > SQL_execute > Data merge > Data Fetch

C1 C2 C3 C4 C5 C6 C7

-------------------------------------------------- -------------------------

1 x 0 abc 1 01-JUN-2021

3 x 0 abc 1 01-JUN-2021

5 x 0 abc 1 01-JUN-2021

7 x 0 abc 1 01-JUN-2021

8 x 0 abc 1 01-JUN-2021

0 x 0 abc 1 01-JUN-2021

2 x 0 abc 1 01-JUN-2021

4 x 0 abc 1 01-JUN-2021

6 x 0 abc 1 01-JUN-2021

9 x 0 abc 1 01-JUN-2021

10 rows selected.

b. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,0) > node1 deferred preparation connection > SQL_execute > Data Fetch

C1 C2 C3 C4 C5 C6 C7

-------------------------------------------------- -------------------------

0 x 0 abc 1 01-JUN-2021

1 row selected.

c. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,4) > node1 deferred preparation connection > SQL_execute > Data Fetch

C1 C2 C3 C4 C5 C6 C7

-------------------------------------------------- -------------------------

4 x 0 abc 1 01-JUN-2021

1 row selected.

d. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,5) > node2 deferred preparation connection > SQL_execute > Data Fetch

C1 C2 C3 C4 C5 C6 C7

-------------------------------------------------- -------------------------

5 x 0 abc 1 01-JUN-2021

1 row selected.

e. (vado_keeper)SQL_prepare > Check key column c1 > SQL_bindParameter(1,9) > node2 deferred preparation connection > SQL_execute > Data Fetch

C1 C2 C3 C4 C5 C6 C7

-------------------------------------------------- -------------------------

9 x 0 abc 1 01-JUN-2021

1 row selected.

Referring to Figure 5, when the main operation is executed on the proxy server node, it is a server side query, and when the main operation is executed on the client, it is a client side query. At this time, the server side query (52 ), the subject of parallel merge processing of data is the proxy server node (2), and in client-side query (51), the subject of data parallel merge processing is the bardo keeper (4) of the client (1).

When connection to a single data node among a plurality of data nodes is required, the client main operation is performed. In this case, the client and a single data node can be connected.

When connection to two or more data nodes among a plurality of data nodes is required, the proxy server main operation is performed. In this case, the proxy server node and the two or more data nodes may be connected.

At this time, in the conventional invention, the connections between the proxy server node and a plurality of data nodes are processed sequentially. On the other hand, in the present invention, the connection of the proxy server node and the two or more data nodes can be processed simultaneously and in parallel.

More specifically, referring to FIG. 5, as an embodiment of the present invention, for example, when a connection to data node #3 (data node #3, 203) is required, the Vado Keeper is the subject and performs a client-side query ( 51) operates, and the bardo keeper and data node 3 (data node #3, 203) are connected. For example, select * from where c1 in(#3) is executed to query the corresponding data.

Referring to FIG. 5, for example, when connection to two or more data nodes numbered 1 to 5 among a plurality of data nodes is required, the proxy server node 2 is the main agent, and the server side query 52 operates. And, the connection between the proxy server node (2) and data nodes #1 to #5 (data nodes #1 - #5, 201, 202, 203, 204, 205) is processed simultaneously and in parallel. For example, select * from where c1 in (#1), select * from where c1 in (#2), select * from where c1 in (#3), select * from where c1 in (#4), select * from where c1 in (#5) is executed in parallel and retrieves the data.

Referring to FIG. 6, the connection in the present invention includes a Deferred Preparation Connection (100). The general way a database connection works is that the Session Manager first establishes a connection with multiple data nodes, and then uses the session when a connection is needed. In the present invention, a DB connection is not first established and then used, but rather when the connection is deemed necessary, for example, when the session manager (6) determines that a connection to a specific data node is necessary after the sharding method is determined. , can be connected to the data node that is the connection target.

As an embodiment of the present invention, for example, referring to Figure 6, when connection 101 with a specific single data node is required, the session manager can connect 101 with the single data node (connection In use, 41). Rather than first establishing a connection with data nodes other than the single data node and then using it, when it is determined that a connection with data nodes other than the single data node is necessary, the session manager 6 connects with the data nodes. A connection can be made. Before establishing a connection with the data nodes, the device waits (102) for connection with the data nodes (waiting for connection (42)).

Accordingly, in the present invention, database resources can be used efficiently because unnecessary connections are not made.

The dual mode sharding method using this dual mode sharding system can be performed on a computer-readable storage medium on which a program is recorded.

A computer-readable storage medium may be any available medium that can be accessed by a general-purpose or special-purpose computer. By way of non-limiting example, computer-readable media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or storage devices containing desired program code in the form of instructions or data structures. and may include any other medium that can be accessed by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Disk and disk, as used herein, include compact disk (CD), laser disk, optical disk, digital versatile disk (DVD), floppy disk, and Blu-ray disk, and disks. ) usually reproduces data magnetically, but discs reproduce data optically with a laser. Combinations of the foregoing should also be included within the scope of computer-readable storage media.

Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention can be implemented in other specific forms without changing its technical idea or essential features. You will be able to understand it. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive.

1: Client
2: Proxy server node
3: Data Node
4: Vado Keeper
5: Application
6: Session Manager
11: Client Sharding
12: a) Request
13: A) Response
21: Proxy server sharding
22: B) Request
23: B) Response
24: c) Request
25: c) Response
31: Shard connections
32: meta connection
41: Connection in use
42: Waiting for connection
51: Client-side query
52: Server-side query
100: Delay ready connection
101: Connect when connection is busy
102: Delay ready connection when waiting for connection
201: Data Node 1 (Data Node #1)
202: Data Node 2 (Data Node #2)
203: Data Node 3 (Data Node #3)
204: Data Node 4 (Data Node #4)
205: Data Node 5 (Data Node #5)
S100: User query analysis
S200: SQL Prepare()
S300: Check key column
S400: SQL BindParameter()
S500: Check Key value

Claims (10)

A dual-mode sharding system that supports client sharding and proxy server sharding,
A client that determines a sharding method, either client sharding or proxy server sharding;
a proxy server node, including key distribution information;
a plurality of data nodes; and
Includes a session manager,
The client receives a SQL query request by user input and transmits a request for provision of key distribution information to the proxy server node,
The proxy server node transmits the key distribution information to the client in response to a request for provision of the key distribution information,
The client determines one of the client sharding and the proxy server sharding based on the key distribution information,
When the client sharding is determined, the session manager connects the client to a single data node among the plurality of data nodes,
When proxy server sharding is determined, the session manager connects two or more data nodes among the plurality of data nodes with the proxy server node,
When performing a connection between the two or more data nodes and the proxy server node, the session manager simultaneously processes the connection between the proxy server node and the two or more data nodes in parallel rather than sequentially,
The session manager does not perform connection with the plurality of data nodes in advance, but performs connection with the data node to be connected after the sharding method is determined. According to claim 1,
The proxy server node operates as a proxy server and further includes shard configuration information,
The data node is a relational data management system (RDMS), a dual mode sharding system that performs the function of storing data. According to claim 1,
The client is a server on which Vado Keeper runs,
When the Vado Keeper receives the SQL query request, the Vado Keeper transmits a request to provide the key distribution information to the proxy server node, and determines whether a connection to a specific data node is necessary based on the key distribution information. A dual mode sharding system that can determine either the client sharding or the proxy server sharding method. According to claim 3,
The Vado Keeper is,
When the client sharding is determined, transmitting a connection request to the data node and receiving a response to the connection request from the data node,
The proxy server node is,
When the proxy server sharding is determined, the bar can receive a connection request from the keeper, check the key value in response to the connection request, and simultaneously request a connection to two or more data nodes among a plurality of data nodes in parallel. A dual-mode sharding system. According to claim 1,
A dual mode sharding system, characterized in that the proxy server node and the data node are built in an independent form (Stand Alone) and can be duplicated. According to claim 1,
The plurality of data nodes are,
Create each data node configuration for each of the plurality of data nodes, create each shard table for each of the plurality of data nodes, and create an index for each shard table. It is configured by creating, setting the sharding method for each shard table, and setting the sharding operation for data input and the input method to the data node,
Dual mode sharding system. A dual-mode sharding method by a dual-mode sharding system supporting client sharding and proxy server sharding,
The dual mode sharding system is,
It includes a client that determines one of client sharding and proxy server sharding, a proxy server node containing key distribution information, a plurality of data nodes, and a session manager,
The dual mode sharding method is,
The client receiving a SQL query request based on a user input, and transmitting a request for providing key distribution information to the proxy server node;
transmitting, by the proxy server node, the key distribution information to the client in response to a request for provision of the key distribution information;
determining, by the client, a sharding method among the client sharding and the proxy server sharding based on the key distribution information; and
When the client sharding is determined, the session manager connects the client to a single data node among the plurality of data nodes,
When the proxy server sharding is determined, the session manager performs a connection between two or more data nodes among the plurality of data nodes and the proxy server node,
In performing the connection between the two or more data nodes and the proxy server node, the session manager simultaneously processes the connection between the proxy server node and the two or more data nodes in parallel rather than sequentially,
The session manager does not perform connection with the plurality of data nodes in advance, but performs connection with the data node to be connected after the sharding method is determined. According to claim 7,
The Vado Keeper running on the client,
Receiving the SQL query request;
transmitting a request to provide the key distribution information to the proxy server node; and
A dual-mode sharding method comprising determining whether a connection to a specific data node is necessary based on the key distribution information and determining one of the client sharding and the proxy server sharding. According to claim 8,
The Vado Keeper is
If the client sharding is decided,
transmitting a connection request to the data node; and
Receiving a response to the connection request from the data node,
The proxy server node is,
When the proxy server sharding is decided,
Receiving a connection request from the Vado Keeper;
A dual mode sharding method comprising checking a key value in response to the connection request and simultaneously requesting connection to two or more data nodes among the plurality of data nodes in parallel. According to claim 7,
The plurality of data nodes are,
Generating each data node configuration for each of the plurality of data nodes, creating each shard table for each of the plurality of data nodes, creating an index for each shard table, and generating each of the plurality of data nodes. A dual-mode sharding method configured by setting the sharding method for the shard table, setting the sharding operation for data input, and the input method to the data node.