KR100614313B1 - Communication structure of node for preserving situation data in network and node and thereof method - Google Patents

Abstract

The present invention has a number corresponding to the number of network lines for each node, the RQ thread for multicasting the node's status information request packet to the other node, the status information transmitted from the RQ thread because there is one for each node When the request packet is received, it consists of a RS thread that generates a response packet containing its own state information and unicasts to the corresponding RQ thread, a state information table storing availability state information between the network and the node, and a network card for network connection. By reducing the number of threads and sockets created, you can reduce system resource waste.

Node, RQ Thread, RS Thread, Status Information

Description

Communication structure of node for preserving situation data in network and node and explanation method             

1 is a diagram showing the configuration of a general cluster solution.

2 is a diagram showing a communication structure of a conventional node.

3 is a diagram showing a state information table of a conventional network node.

4 is a flowchart illustrating the operation of a conventional R thread.

5 is a flowchart showing the operation of a conventional S thread.

6 is a flowchart showing the operation of a conventional E thread.

7 illustrates a communication structure of a node according to an embodiment of the present invention.

8 is a view showing a state information table according to an embodiment of the present invention.

9 is a flowchart illustrating the operation of an RQ thread according to an embodiment of the present invention.

10 is a flowchart illustrating operation of an RS thread according to an embodiment of the present invention.

<Explanation of symbols for the main parts of the drawings>

700: node 702, 810: RQ thread

704, 820: RS thread 701: socket

706: network card 800: status information table

The present invention relates to a communication structure of a node for maintaining network and node state information for minimizing the use of system resources when attempting to maintain network and node availability state information in real time in a multi-network, multi-node cluster environment. It is about how.

Telecommunication systems typically consist of two or more multi-process systems, shared disks, a heartbeat network and telecommunications software. The information communication software is an important component that analyzes the state of the system, delivers the information to other nodes, and takes over the service performed by the node to another node when a node fails.

The information and communication system is a system that recognizes the failure and takes over the service of the failed system and guarantees the continuity of the service when a hardware or software failure occurs in the system providing the service.

Recently, with the development of information technology, the performance and safety of hardware have been rapidly improved, but the software has caused a lot of failures due to the relatively increased complexity of the software.

Thus, the ability to monitor and overcome software failures in information and communication systems is becoming increasingly important.

1 is a diagram illustrating a configuration of a general cluster solution.

Referring to FIG. 1, the cluster solution determines whether a network and a node have failed and transmits cluster control and status information.

Such a cluster includes a plurality of nodes (100a, 100b, 100c, hereinafter 100), a heartbeat network 120, a hub 130, and a public network for monitoring and exchanging information with each other. 110).

The cluster solution implements optimized communication and failure determination by selecting an available network and node 100 using the state information stored in each node 100.

Detailed description of the node 100 is shown in FIG. 2.

2 is a diagram illustrating a communication structure of a conventional node.

Referring to FIG. 2, node 200 includes R threads 202a and 202b, S threads 206a and 206b, E threads 204a and 204b, network cards 208a and 208b, and a state information table (not shown). It includes.

The R threads 202a and 202b are threads that receive state information multicasted by the S threads of other nodes.

The E threads 204a and 204b are threads for updating their own state information.

The S threads 206a and 206b transmit their state information to the R thread of another node.

The R threads 202a and 202b and the S threads 206a and 206b each create one socket 201 in the corresponding network line. Therefore, it can be seen that when there are a plurality of network cards, threads and sockets are created in proportion to the number of network cards.

Description of each of the R threads 202a and 202b, the S threads 206a and 206b, and the E threads 204a and 204b will be described with reference to FIGS. 4 to 6.

3 is a view showing a state information table of a conventional node.

Referring to FIG. 3, the state information table 300 is a table that stores availability state information between a network and a node. Therefore, the state information table 300 stores state information of each node and its own state information transmitted through a network line connected to the corresponding node.

In addition, the value of the state information table 300 is set to have a maximum and minimum values that can be stored, and the minimum value represents a state in which communication with the node is not possible through the corresponding network.

The R thread 310, the S thread 330, and the E thread 320 of each node grasp availability status information between nodes through a network line.

The R thread 310 receives the node state information multicasted by another node and records it in the corresponding area of the state information table.

The S thread 330 multicasts node state information to another node.

The thread E 320 updates the node's own state information.

4 is a flowchart illustrating the operation of a conventional R thread.

Referring to FIG. 4, the R thread creates a socket for communication and binds to a specific port (S400). Then, the R thread waits for receiving node state information multicasted by the S thread of another node (S402), and determines whether node state information is received (S404).

When the node state information is received as a result of the determination in step S404, the R thread analyzes the received state information and displays an available state of the corresponding node in the state information table (S406). That is, when state information is received, the R thread determines that the node is in an available state and increases the value of the corresponding area of the state information table.

If the node state information is not received as a result of the determination in step S404, the R thread determines whether or not the predetermined state of the state information has elapsed (S408).

If the state information is not received within a predetermined time period as a result of the determination of step S408, the R thread displays the unavailable state of the corresponding node in the state information table (S410). That is, the R thread determines that a node that has not received state information within a predetermined time period is not normal, and displays an unavailable state on the corresponding node of the state information table.

If the state information is received within a predetermined time as a result of the determination in step S408, the R thread performs step S406.

5 is a flowchart illustrating the operation of a conventional S thread.

Referring to FIG. 5, the S thread creates a socket for communicating and binds a corresponding network line (S500). The network line refers to a local address of the corresponding network line.

After performing step S500, the S thread collects node information corresponding to the corresponding network line from the state information table and multicasts it to an R thread of another node (S502).

After performing step S502, the thread S stops operation for a predetermined time (S504), and then repeats step S502.

6 is a flowchart illustrating the operation of a conventional E thread.

Referring to FIG. 6, the E thread calculates state information of another node on a corresponding network line in a state information table to obtain state information of the corresponding node (S600). Here, the operation may be an or operation. That is, the E thread may determine that its own node is in an available state if one of the other nodes on the network line is available in the state information table, and obtain a state information value of the corresponding node.

After performing step S600, the E thread stores the obtained state information value in the state information table (S602), and decreases all node state information values on the corresponding network line in the state information table (S604). By reducing all node state information values on the corresponding network line, a failed network and a node can be detected.

Then, the E thread stops the operation for a predetermined time (S606), and repeats from step S600.

However, as described above, since the number of threads and sockets generated in the related art increases in proportion to the number of network cards, system resources are frequently used.

In addition, the R and E threads simultaneously write to the same area of the state information table, which is a shared variable, which must be controlled by using a synchronization mechanism, which not only complicates the algorithm itself but also causes additional load during algorithm execution.

In addition, the multicast communication method used is a UDP method that does not provide flow control, and thus lacks stability.

Accordingly, an object of the present invention is to provide a node communication structure and method for maintaining state information of a network and a node capable of maintaining network and node availability state information in real time while minimizing the use of system resources.

According to an aspect of the present invention to achieve the above objects, there is a number corresponding to the number of network lines for each node, RQ thread for multicasting the node's status information request packet to another node, 1 for each node When there is a dog and a status information request packet transmitted from the RQ thread is received, a RS packet unicast to the corresponding RQ thread by generating a response packet including its own status information and storing availability status information between the network and the node. A communication structure of a node is provided which comprises an information table and a network card for network connection.

The RQ thread or the RS thread each creates one socket for communication.

The state information table stores state information of each node and its own state information transmitted through a network line.

According to another aspect of the present invention, in the operation of the RQ thread to maintain state information of the network and the node, create a socket for performing communication and bind to the corresponding network line, initialize a temporary variable to store the response packet Multicasting the status information request packet to the RS thread of another node, and receiving the status information response packet from the RS thread of the other node, storing the status information in the temporary variable, and when a predetermined time elapses, A method of maintaining state information of a network and a node is provided, wherein the state information stored in a temporary variable is reflected in a state information table.

If a status information response packet is not received within a predetermined number of times from an RS thread of another node, it is determined that a failure has occurred in the corresponding node, and an unavailable status is displayed in the status information table.

The RQ thread waits on the created socket to receive a status response packet from the RS thread of another node.

Reflecting the state information stored in the temporary variable in the state information table increases the value of the corresponding region in the state information table for the node where the response packet is received, and decreases the value of the remaining nodes.

According to another aspect of the present invention, in the operation of the RS thread for maintaining the state information of the network and the node, create a socket for communication and bind to a specific port, when the node state information request packet is received, the counterpart A method of maintaining network and node state information is provided, which stores an address and a port and unicasts its own node state information to a corresponding counterpart.

The RS thread waits on the created socket to receive a state information request packet multicasted by an RQ thread of another node.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

7 is a diagram showing a communication structure of a node according to an embodiment of the present invention.

Referring to FIG. 7, node 700 includes RQ threads 702a and 702b, RS thread 704, network cards 706a and 706b, and a state information table (not shown).

The RQ threads 702a and 702b request status information from an RS thread of another node, and the RS thread 704 transmits its status information to the RQ thread of another node.

Here, the RQ threads 702a and 702b are proportional to the number of network lines, and the RS thread 704 is one for each node.

The RQ threads 702a and 702b and the RS thread 704 create one socket 701a and 701b for communication.

The sockets 701a and 701b of the RQ threads 702a and 702b bind to the local address of the corresponding network line and multicast the node state information request packet to another node.

The socket of the RS thread 704 binds to a specific port, receives a node state information request packet multicasted by another node, and generates and processes a response packet.

The RS threads 702a and 702b and the RQ thread 704 will be described in detail with reference to FIGS. 9 and 10.

8 is a diagram illustrating a state information table according to an embodiment of the present invention.

Referring to FIG. 8, the state information table 800 is a table that stores availability state information between a network and a node. Therefore, the state information table 800 stores state information of each node and its own state information transmitted through a network line connected to the corresponding node.

In addition, the value of the state information table 800 is set to have a maximum and minimum values that can be stored, and the minimum value represents a state in which communication with a node is not possible through a corresponding network.

The RQ thread 810 and the RS thread 820 of each node grasp availability status information between nodes through a network line.

The RQ thread 810 multicasts a node status information request packet to an RS thread of another node to grasp connection status information with another node on a network line. In addition, the RQ thread 810 receives and analyzes a unicast response packet transmitted from each node and stores the result in a state information table.

The RS thread 820 operates as one thread regardless of a network line, and when a status information request packet multicasted by an RQ thread of another node is received, the RS thread 820 sends a response packet through the received network line. Unicast

Here, when the access patterns of the state information tables of the RQ thread 810 and the RS thread 820 are observed, it can be seen that a plurality of threads do not simultaneously write to the same area.

9 is a flowchart illustrating the operation of an RQ thread according to an embodiment of the present invention.

9, the RQ thread creates a socket for performing communication and binds to a corresponding network line (S900). Then, the RQ thread initializes a temporary variable to store the response packet (S902), and multicasts the status information request packet to the RS thread of another node (S904).

Then, the RQ thread determines whether state information is received from the RS thread of another node (S906). That is, the RQ thread waits on a socket to receive a node status information response packet unicast by the RS thread.

As a result of the determination of step S906, when state information is received from the RS thread of another node, the RQ thread stores the received state information in a temporary variable (S908), and determines whether a predetermined predetermined time has elapsed (S910). ).

If a predetermined time elapses as a result of the determination in step S910, the RQ thread reflects the value stored in the temporary variable in the state information table (S912). That is, the RQ thread increases the value of the corresponding region of the state information table for the node that receives the response packet and decreases the value for the remaining nodes.

If the state information is not received from the RS thread of another node as a result of the determination of step S906, the RQ thread determines whether the state information is received within a predetermined number of times (S914).

If the state information is received within a predetermined number of times as a result of the determination of step S914, the RQ thread performs step S908.

If the state information is not received within a predetermined number of times as a result of the determination of step S914, the RQ thread displays the unavailable state of the corresponding node in the state information table (S916).

10 is a flowchart illustrating the operation of an RS thread according to an embodiment of the present invention.

Referring to FIG. 10, the RS thread creates a socket and binds to a specific port (S1000). The RS thread must not bind to a specific address in order to receive multicast requests from all network lines.

After performing step S1000, the RS thread determines whether a node status information request packet is received (S1002). That is, the RS thread waits on a socket to receive a node state information request packet multicasted by the RQ thread.

As a result of the determination of step S1002, when the node status information request packet is received, the RS thread stores the address and port of the counterpart (S1004), and unicasts its node state information to the counterpart (S1006).

The present invention is not limited to the above embodiments, and many variations are possible by those skilled in the art within the spirit of the present invention.

As described above, according to the present invention, since the number of generated threads and sockets is reduced, it is possible to provide a communication structure and a method of a node for maintaining state information of a network and a node, which can reduce system resource waste.

In addition, according to the present invention, since each thread does not simultaneously perform a write operation on the same area of the state information table, which is a shared variable, it does not use a synchronization mechanism and does not use a synchronization mechanism, thereby simplifying a state information maintenance algorithm and It is possible to provide a communication structure and a method of a node for maintaining the state information of the network and the node can remove the load.

In addition, according to the present invention, by using the flow control method of the request and response structure for transmitting and receiving the node status information, it is possible to maintain the status information of the network and the node to compensate for the disadvantages of the UDP-based communication scheme and improve the reliability of the status information transmission. It is possible to provide a communication structure of a node and a method thereof.

Claims (9)

In the communication structure of a node for maintaining network and node state information, An RQ thread for each node corresponding to the number of network lines and multicasting a node's status information request packet to another node; When there is one for each node and receives the multicast status information request packet from the RQ thread, the RS thread generates a response packet including its own status information and unicasts to the corresponding RQ thread; A status information table for storing availability status information between the network and the nodes; and Network card for network access Communication structure of the node comprising a. The method of claim 1, The RQ thread or the RS thread each creates a socket for communication. The method of claim 1, The state information table stores the state information of each node and its own state information transmitted through a network line. In the operation of RQ threads to maintain network and node state information, Creating a socket for performing communication and binding to a corresponding network line; Initializing a temporary variable to store a response packet, and multicasting a status information request packet to an RS thread of another node; If the status information response packet is received from an RS thread of another node, storing the status information in the temporary variable; Reflecting state information stored in the temporary variable in a state information table when a predetermined predetermined time elapses Method of maintaining state information of the network and the node comprising a. The method of claim 4, wherein If the status information response packet is not received within a predetermined number of times from the RS thread of the other node, determining that a failure has occurred in the corresponding node, and displaying an unavailable state in the status information table. To maintain state information for nodes and nodes. The method of claim 4, wherein And the RQ thread waits on the created socket to receive a status response packet from an RS thread of another node. The method of claim 4, wherein Reflecting the state information stored in the temporary variable in the state information table includes increasing the value of the corresponding region in the state information table for the node where the response packet is received, and decreasing the value of the remaining nodes. How to maintain state information for a node. In the operation of RS threads to maintain network and node state information, Creating a socket for communication and binding to a specific port; When the node status information request packet is received, storing an address and a port of the corresponding counterpart; Unicasting its node state information to its counterpart Method of maintaining state information of the network and the node comprising a. The method of claim 8, And the RS thread waits on the created socket to receive a state information request packet multicasted by an RQ thread of another node.

Images