KR100361076B1 - Method of mimory sharing and duplicate server system thereof - Google Patents

Abstract

The present invention relates to a server redundancy system and method, and more particularly, to a method and system that can share and access memory contents in a fast time between a plurality of hosts connected to a network.

According to the present invention, a plurality of hosts having a reflective memory updated in real time are connected to each other through a reflective memory data transfer means to form a network, and each reflective host includes a reflective memory library. After allocating a local data area that can be referenced by networked hosts in the memory area of the creative memory, the user can write, store, or read a message or data in the allocated local data area by using a reflective memory library. Connected hosts have the same effect as reading data from their main memory. As a result, it is possible to guarantee a fast and constant network transmission speed regardless of the amount of network traffic in communication for data sharing between the memory of the redundant hosts.

Description

MEMORY OF MIMORY SHARING AND DUPLICATE SERVER SYSTEM THEREOF}

The present invention relates to a server redundancy method and system, and more particularly, to a memory sharing method and system in which redundant hosts connected to a network share the same data in real time to access a host.

In recent years, as network technology has evolved, multimedia networking technology has been introduced, which allows users to connect to a network in real time, such as video conferencing, distance learning, and network games, to transmit and share voice and video data with each other. Is going.

However, multimedia networking applications are demanding high-speed transmission of a large amount of data generated from video, audio, and information processing, and require a certain transmission speed higher than a required level despite heavy network traffic. In addition, it is required that a large number of hosts connected to a network share data in real time at the same time. The voice mail system (VMS) or unified message system (UMS) is basically stopped. We aim for a service that is not available. To this end, two hosts should be configured to monitor each other, and in the event of an abnormality or failure, a switchover must be performed to recover the service in a short time. It consists of a (standby) structure. In other words, one of the two servers is in an active state and the other is in a standby state. However, there is no difference in that both the server in the active state and the server in the standby state are booted and the operating system (OS) is operating, but from a functional point of view, the server is in the active state. This is different from the MCU / MFU in standby state because the process required only for the MCU's main control unit / multi-function unit (MFU) is started and provides services. Monitor the MCU / MFU of the opponent's active state. Of course, the standby MCU / MFU must also have processes running for monitoring the active MCU / MFU; however, most of the processes on the standby server side are switched off when they detect that the active MCU / MFU is down. Immediately afterwards, it is activated and performs the function as an active one. In general, in a redundant server system, the main control unit (MCU) manages subscriber information, service information, and a system / service configuration database. In an active / standby switchover, a database management server (DBMS) Safe transfer of) is required. In addition, a rapid transfer process is required for active / standby switching. Furthermore, a synchronized memory sharing method is required in which a standby server, which is transferred in real time, recognizes data contents recorded by an active MCU / MFU. do.

Accordingly, a first object of the present invention is to provide a communication method and system in which an active server and a standby server can share data contents in synchronization in real time in a redundant server system that is networked.

A second object of the present invention is to provide a communication method and system that can be switched quickly and stably in an active / standby switchover in a networked redundant server system in addition to the first object.

1 is a block diagram of a host computer constituting a redundant server system according to the present invention.

Fig. 2A is a diagram showing a memory area of a reflective memory included in a host computer constituting a redundant server system according to the present invention.

Figure 2B illustrates one embodiment of allocating a memory region of a reflective memory in accordance with the present invention.

3A is a system configuration diagram of a redundant server system according to a first embodiment of the present invention.

Figure 3b illustrates an initial operation for server redundancy according to the first embodiment of the present invention.

Figure 3C is a flow diagram illustrating the workflow of a function to initialize a reflective memory device in a reflective memory library in accordance with the present invention.

FIG. 3D is a flow diagram illustrating a flow of allocating a local data area to a reflective memory using a reflective memory library in accordance with the present invention. FIG.

3E is a flow diagram illustrating a workflow for storing or reading data in a local data area allocated to the reflective memory using the reflective memory library in accordance with the present invention.

3F is a flowchart showing a server redundancy method according to the first embodiment of the present invention.

4A is a system configuration diagram of a redundant server system according to a second embodiment of the present invention.

Fig. 4B shows a preferred method of allocating a memory area of a reflective memory that can be used in a redundant server system according to a second embodiment of the present invention.

FIG. 4C illustrates another embodiment of the memory area allocation of the reflective memory that can be used in the redundant server system according to the second embodiment of the present invention. FIG.

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

100, 200: host computer

110, 210: Microprocessor

120, 220: main memory

130, 230: Local storage means (hard disk)

140 and 240: reflective memory (RFM)

300: reflective memory data transmission and reception means (optical cable)

400: memory area of reflective memory

410: Reflective memory device driver usage area

420: base register area

430 meta database area

440: user data area

480: memory address of reflective memory

In order to achieve the above object, the present invention is directed to a &quot; redundant system composed of a plurality of hosts operating in an active or standby state and connected to each other via a network, wherein each host has a data transmission / reception port, A reflective memory (RFM) having a function of storing or reading the data; and connected to a data transmission / reception port of the reflective memory provided in the host and stored in the plurality of reflective memories Reflective memory data transfer means for synchronizing the contents in real time in the same manner, wherein data generated by a process running in an active host among the plurality of hosts includes the reflective memory and the reflective memory data of the active host. Synchronously connected by transmission means By being written with the reflective memory of a tan by the host provides a redundant system ", characterized in that the shared memory contents during active / standby switching.

In order to achieve another object of the present invention, "In a method of performing data communication for sharing data generated by a running process with a standby host in an active host of a redundant system consisting of a plurality of hosts operating in an active or standby mode Creating a server process or client process on the host; Allocating a local data area to a memory area of the reflective memory mounted in the host by using a reflective memory library; The client process writing a message to the allocated local data area using a reflective memory library; The server process reads a message of the client process from the allocated local data area using a reflective memory library and writes a response message using the reflective memory library; And reading, by the client process, a response message of the server process using the reflective memory library in the allocated local data area.

Hereinafter, a preferred embodiment of a real-time communication method of a redundant server and a system using the same according to the present invention will be described in detail with reference to FIGS. 1 to 4.

1 is a block diagram of a host computer constituting a redundant server system according to the present invention. Referring to FIG. 1, the host computer 100 constituting the network communication system according to the present invention includes a microprocessor 110 for generating various processes and a main memory 120 for storing data generated by the process. The hard disk 130 is provided as a local storage means for storing data or data to be transmitted.

In addition, the host computer 100 constituting the redundant server system according to the present invention includes a reflective memory (RFM; 140) in addition to the microprocessor (110), the main memory (120), and the hard disk (130). Characterized in that.

The RFM 140 used as a preferred embodiment of the present invention may be mounted and used in a PCI slot of the host computer 100, and may be connected to the data transmission / reception means 300 having a transmission / reception port for transmitting and receiving data. . In addition, by using the data transmission and reception means 300 connected to the transmission and reception port, it may have the same memory structure as the RFM of another standby server connected to the RFM 140 of the active server in hardware. Two MCU / MFUs (main control unit; multi-function unit) are configured to monitor each other, and in the event of an abnormality or failure, quickly switch to recover the service.The active side MCU / MFU is configured as pnr.tab. All processes described in the configuration table will be started. On the other hand, the standby MCU / MFU is characterized in that all processes other than the process and resource (PNR) and dual guardian process (DNP) does not work. PNR according to the present invention is a process that manages the process, MCU / MFU must be started when it first boots up. Therefore, when the MCU / MFU is booted and is in a running state, the PNR process must be started. That is, the PNR process according to the present invention is the first process management table (pnr.tab) once it is started. Read). In other words, it obtains information about the process's operation from the specified configuration file and starts the process. When the started process (which becomes a child process for the PNR) is normal or abnormal or terminates, It finds and starts the process again by referring to the setting value in the process management table. That is, the PNR according to the present invention is the parent process of all processes in the MCU / MFU. Everybody dies. The PNR also monitors and manages the status of network interface cards (NICs) installed in its systems for network redundancy. Here, for complete redundancy, the information managed by each process must also be duplicated. If this information is stored in shared memory or local memory in main memory, a sudden MCU / MFU switchover would have occurred. In other words, when the MCU / MFU transition due to automatic switching occurs, there is no way to find the existing information from the newly active MCU / MFU. In order to solve this problem, the present invention uses a reflective memory (RFM). The RFM according to the present invention is a memory board which is additionally mounted in addition to the main memory of the MCU / MFU. Currently, since the bus scheme adopted by the MCU / MFU is a PCI scheme, the RFM according to the present invention may also use the PCI bus scheme. The RFM boat for PCI according to the present invention is installed in each of the two MCU / MFU in the active and standby state, and can be connected by a fiber optic cable between the two memory boards. According to a preferred embodiment of the present invention, MCU / Processes in the MFU are written to either side of the RFM if either side writes, so that the contents of both RFMs are always consistent and synchronized. The advantage of data transfer is that no software operation is involved. That is, when data is written to a register of one of two RFM nodes, the same contents are automatically transferred and written to the same register of the opposite node through the optical cable. Therefore, in case of applying this, synchronization of memory contents can be realized in near real time, which can be usefully used for dualizing MCU / MFU.

Figure 2a is a view showing the structure of the memory of the RFM mounted to the host computer constituting the redundant server system according to the present invention. Referring to FIG. 2A, a memory area 400 and a memory address 480 of the RFM 140 are shown.

The memory area 400 of the RFM 140 may include an RFM device driver use area 410, a base register area 420, a metadatabase area 430, and a user data area 440. Can be.

In a preferred embodiment of the present invention, the RFM device driver usage area 410 may be a 64 byte area from 0x0000 to 0x0040 of the memory area 400 of the RFM 140, and the RFM device driver may be an RFM. Data used for driving the 140 may be stored.

The base register area 420 of the RFM 140 according to the present invention is an area from the RFM device driver area 410 to the address 0x0044, and may store a start address of a user available area of the memory area of the RFM 140. have.

In addition, the metadata area 430 is an area occupying a size of 0x04B4 bytes after the base register area 420 and may store information of a local data area allocated to data to be stored in the user data area 440. In a preferred embodiment of the present invention, the local data area information includes a reflective memory address identifier (RfmID) of the local data area allocated for storing the corresponding data in the RFM 140 and the size of the allocated local data area. memory size; RfmSize), and a delimiter to distinguish the data.

As a preferred embodiment of the RfmKey according to the present invention, a process generated in the microprocessor 110 may determine when the data is stored in the RFM 140, and is allocated from '0' to the memory area 400 of the RFM 140. Can be determined between integers representing the maximum number of local data areas that can be.

2B illustrates an embodiment of allocating a memory area of an RFM according to the present invention. Referring to FIG. 2B, addresses 0x04f8 to 0x09AC in the user data area 440 are already allocated local data areas 441 for data storage, and after address 0x09AD, they are unallocated areas 442. Accordingly, the base register region 420 stores a value of 0x09AD, which is a start address 425 of an available memory region of the RFM 140 memory region 400.

In addition, the meta database area 430 stores RfmKey 451, RfmID 452, and RfmSize 453, which are local data area information 450 of the local data area 441, which is already allocated for data. That is, the RfmID 452 value stores 0x04F8, which is the start address of the local data area 441 allocated for data storage, and the RfmSize 452 value, 0x04B4, which is the size of the local data area 441. The value is stored.

Accordingly, if only the RfmKey 451 value determined by the process generated by the microprocessor 110 is known, the metadata area 430 is searched with the RfmKey 451 value to obtain the corresponding RfmID 452 and RfmSize 453 to obtain the RFM. Data may be stored or read in the local data area 441 of 140.

3A is a system configuration diagram of a network communication system according to a first embodiment of the present invention. Referring to FIG. 3A, the first host 100 and the second host 200 constituting the network communication system according to the first embodiment of the present invention are equipped with RFMs 140 and 240 in PCI slots, respectively. Since the data transmission and reception means 300 are connected to the transmission and reception ports of the 140 and 240, the connection is performed.

As a preferred embodiment of the present invention, the data transmission and reception means 300 may be an optical cable 300, the optical cable 300 used can transmit data at a high transmission rate of less than nanoseconds (nano-second) . Therefore, data written to the RFM 140 mounted on the first host 100 is written and automatically transmitted to the RFM 240 mounted on the second host 200 through the optical cable 300. The data of the RFMs 140 and 240 can be synchronized in near real time.

3B is a diagram illustrating an initial operation for server redundancy according to the first embodiment of the present invention. Referring to FIG. 3B, a first host 100 and a second host 200 are illustrated, and RFMs 140 and 240 mounted on each host 100 and 200 are connected through an optical cable 300. have. In a preferred embodiment according to the present invention, the first host 100 may be a server and the second host 200 may be a client.

According to a preferred embodiment of the present invention, the microprocessors 110 and 210 of each host 100 and 200 generate a server process 115 and a client process 215, respectively, and use an RFM library using the RFM library. And a RfmKey value at 240, and allocates a local data area for the corresponding RfmKey in the memory area 400 of the RFM 140 and 240. A process and resource manager (PNR) that is always started when the MCU / MFU is first booted. The process will start a duplication guardian process (DGP). The DGP process according to the present invention periodically communicates with the other party's DGP via a heart beat network, and detects whether the other party is transferred so that the transfer operation is performed. In other words, DGP is in charge of the monitoring function for redundancy between active / standby MFU, and can communicate with other DGP periodically through heartbeat network that is separate from service network. The RFM library can be used to read and write data to the RFM according to the present invention. If two or more processes need to access the same RFM area, use a semaphore to lock them.Each process is allocated a certain area within the RFM and shared memory ( You can get a pointer to the allocated area of the RFM, like using shared memory.

In a preferred embodiment of the present invention, the RFM library includes a function of opening the RFM 140, 240 device (RfmOpen), a function of closing the RFM 140, 240 device (RfmClose), an RFM 140, 240 device. Function to initialize (RfmInit), get RfmID value of memory area of RFM (140, 240) with RfmKey value set by process (RfmGet), RFM (140, with RfmID value as starting address) A function that obtains a pointer on main memory 120 pointing to a memory area of 240 (RfmAttach), sets a semaphore lock state for a specific RfmKey value, and sets the semaphore lock state for a specific RfmKey value. The RfmSemUnlock function may include a function of forcibly synchronizing the contents of data stored in the RFMs 140 and 240 (RfmSync).

In a preferred embodiment of the present invention, the RfmSync function writes data stored in the memory area 400 of the RFM 140 mounted on the server 100 to the RFM 140 once again, so that the respective RFMs 140, 240 The data stored in the memory area 400 may be forcibly synchronized.

In a preferred embodiment of the present invention, the RfmSemLock function and the RfmSemUnlock function are functions for changing the state of the semaphore, so that the microprocessors 110 and 210 of each host 100 and 200 simultaneously share the same memory area of the RFM 140 and 240. It can be used to prevent access and to generate event signals when storing or reading data.

That is, before the first host 100 stores or reads data in the RFMs 140 and 240, the RfmSemLock function is executed to set a semaphore for the RfmKey of the corresponding data and then store the data. The second host 200 is prevented from accessing the same memory area of the RFMs 140 and 240 with the corresponding RfmKey value so that no data error occurs.

In addition, when the first host 100 finishes storing or reading data, the RfmSemUnlock function is executed to release the semaphore lock state, and the second host 200 accesses a corresponding area of the RFMs 140 and 240. Make it possible. In addition, the second host 200 may detect that the lock state of the semaphore for the specific RfmKey changes, and recognize that the first host has accessed the RFMs 140 and 240.

3C is a flow diagram illustrating the workflow of a function of initializing an RFM device in an RFM library in accordance with the present invention. Referring to FIG. 3C, the RfmInit function for initializing the RFM 140, 240 device may include a memory area 400 of the RFM 140, 240 and a memory area of the main memory 120, 220 of each host 100, 200. (Step S501), the base register area 420 is allocated to the RFMs 140 and 240 (step S502).

Subsequently, a semaphore is generated for the RFM 140, 240 memory area, and an initial value is set (step S503). After the generation of the semaphore and the initial value setting are completed, the meta database area 430 is allocated (step S504), and the start address of the user data area 440 available to the user after the meta database area 430 is assigned to the base register ( 420) update to a value (step S505).

3d is a flowchart illustrating a flow of allocating a local data area to an RFM using an RfmKey value and an RFM library according to the present invention. Referring to FIG. 3D, the server process 115 and the client process 215 of each host 100, 200 execute the RfmOpen function to open the RFM 140, 240 device (step S601).

Further, the RfmInit function is executed to allocate the base register 420 and the meta database area 430 (step S602). The server process 115 or client process 215 then determines the RfmKey value, reads from the base register 420 the start address of the available area of the user data area 440 using the RFM library, and reads the RfmSize value. As much as the local data area for the RfmKey value is reserved on the user data area 440 (step S603).

Subsequently, local data area information such as RfmKey, RfmID, and RfmSize is stored in the meta database area 430 using the RFM library, with the start address of the usable area read out from the base register 420 as RfmID (step Sfm). S604).

3E is a flowchart illustrating a data storage or reading method using an RFM library according to the present invention. Referring to FIG. 3E, the RfmGet function is executed to read the RfmID by using the RfmKey value (step S701), and the RfmID is executed to execute the RfmAttach function to obtain a pointer in the main memory indicating the region having the RfmID as the starting address. (Step S702).

Subsequently, the semaphore for the RfmKey value is executed by executing the RfmSemLock function to set the semaphore lock state (step S703), and the data is stored or read in the area pointed to by the pointer obtained by the RfmAttach function (step S704), and the RfmKey value is again obtained. The RfmSemUnlock function is executed to release the semaphore lock state (step S705).

According to the present invention, by using the RfmGet function and the RfmAttach function included in the RFM library, the microprocessors 110 and 210 of each host 100 and 200 store the RFM as if the data is stored or read in main memory. Data can be stored or read in the memory area of the memory.

As a preferred embodiment of the present invention, a predetermined RfmKey value may be used for storing or reading specific data values. That is, the RfmKey value for storing or reading data in the base register area 420 may use '0', and the RfmKey value for storing and reading data in the meta database area 430 may read '1'. Can be used.

3F is a flowchart illustrating a communication method of a redundant server according to the first embodiment of the present invention. Referring to FIG. 3F, when the server process 115 and the client process 215 open and initialize the RFMs 140 and 240 (step S801), the server process 115 determines the RfmKey value and the data storage area. To store the data area information in the meta database area 430 (step S802).

Subsequently, the client process 215 is notified of the RfmKey value from the server process 115, searches the meta database area 430 with the notified RfmKey value, and obtains an RfmID. Further, with the obtained RfmID, a pointer on main memory 120, 220 indicating the allocated data area is obtained (step S803). The data request message is written to the RFMs 140 and 240 using the pointer obtained as the RfmID (step S804).

The server process 115 recognizes the change in the semaphore lock state according to the writing of the data request message, reads the data request message of the client process 215 from the RFM 140 (step S805), and corresponds to the data request message. The data to be read is read from the hard disk 130 of the first host 100 and written to the RFM 140 of the first host 100 (step S806).

In addition, the client process 215 is a semaphore lock that occurs because the server process 115 reads data from the hard disk 130 of the first host 100 and writes the data to the RFM 140 of the first host 100. Recognizing a change in state, the requested data is read from the RFM 240 of the second host 200 updated in real time and stored in the hard disk 230.

Meanwhile, in the step in which the client process 215 is notified of the RfmKey value from the server process 115, the RFM 140, 240 may assign the RfmKey value to the client process 215 using the selected RfmKey value. ), And the client process 215 executes the RfmGet function and the RfmAttach function with the selected RfmKey value to read the RfmKey value assigned to the client process 215 of the second host 200. Can be.

In addition, the step of reading data from the hard disk 130 of the first host 100 and writing the data to the RFMs 140 and 240 (step S806) is performed by the first host 100 requesting a message from the second host 200. This can be done by writing a reply message to RFM 140. When the second host 200 reads the data written in the RFM 240 and stores the data in the hard disk 230 (step S807), the second host 200 transmits the reply message of the first host 100 to the RFM. It can be substituted by reading from 140 and 240.

In addition, as a preferred embodiment of the present invention, the step of reading data from the hard disk 130 of the first host 100 to write to the RFM 140 (step S806) and the RFM 240 of the second host 200 In step S807, the data is read out and stored in the hard disk 230 (step S807), and may be repeated once or more depending on the size of the data.

According to the present invention, the RFMs 140 and 240 that are updated in real time can immediately read data in the remote host from the RFM that they are equipped with in a short time without delay by the circuit. In addition, by using the RFM library according to the present invention, the user can obtain an effect such as storing or reading data on the main memory.

4A is a block diagram of a multiplexing server system according to a second embodiment of the present invention. Referring to FIG. 4A, each host 200 connected to the optical cable 300 in a ring type with the first host 100 is illustrated. In a preferred embodiment of the present invention, the first host 100 may be a server 100, and each host 200 may be a client host 200.

The server 100 and the client host 200 according to the present invention open and initialize the RFMs 140 and 240 using the RFM library. In addition, the client host 200 is allocated a memory area 400 of the RFMs 140 and 240 for network communication using the RfmKey value assigned by the server 100. That is, each host 100 or 200 may receive a local data area in the memory area 400 of the RFMs 140 and 240 so as to exchange a message or data.

According to the second embodiment of the present invention, each client host 200 may guarantee a constant network transmission speed so that each client host 200 may obtain data from the server 100 in a short time through the RFMs 140 and 240 updated in real time. .

4B is a diagram illustrating an RFM memory area allocation method that can be applied to a network communication system according to a second embodiment of the present invention. Referring to FIG. 4B, the memory area 400 and the memory address 480 of the RFMs 140 and 240 are shown. The memory area 400 of the RFMs 140 and 240 may be divided into an RFM device driver use area 410, a base register area 420, a meta database area 430, a user data area 440, and the like.

According to the present invention, a plurality of client hosts 200 accessing the server 100 receive different RfmKey values 901, 902, and 903 from the server 100, and thus, different local data areas 911, 912, and 913. Can be assigned. Accordingly, each client host 200 may perform network communication with the server 100 independently of each other.

4C is a diagram illustrating another embodiment of an RFM memory area allocation method that may be used in a network communication system according to a second embodiment of the present invention. Referring to FIG. 4C, the memory area 400 and the memory address 480 of the RFMs 140 and 240 are shown. The memory area 400 of the RFMs 140 and 240 may be divided into an RFM device driver use area 410, a base register area 420, a meta database area 430, a user data area 440, and the like.

As a preferred embodiment of the present invention, a plurality of client hosts 200 accessing the server 100 may be assigned the same RfmKey 904 value from the server 100 and may be assigned the same local data area 914. Accordingly, each client host 200 may simultaneously share data with each other by referring to the same local data area 914 among the memory areas 400 of the RFMs 140 and 240 mounted in each host 200. That is, when the data of the local data area 914 is changed by the server 100, all the client hosts 200 directly refer to the same data in the RFM 200 to which they are mounted.

The foregoing has outlined rather broadly the features and technical advantages of the present invention to better understand the claims of the invention which will be described later. Additional features and advantages that make up the claims of the present invention will be described below. It should be appreciated by those skilled in the art that the conception and specific embodiments of the invention disclosed may be readily used as a basis for designing or modifying other structures for carrying out similar purposes to the invention.

In addition, the inventive concepts and embodiments disclosed herein may be used by those skilled in the art as a basis for modifying or designing other structures for carrying out the same purposes of the present invention. In addition, such modifications or altered equivalent structures by those skilled in the art may be variously changed, substituted, and changed without departing from the spirit or scope of the invention described in the claims.

As described above, the present invention provides a unique memory area having a reflective memory in which a plurality of host computers constituting the multiplexing server system are updated in real time, and which can be referred to between host computers on the reflective memory. By allocating and writing a message or data in the allocated memory area, it is possible to send and receive a message or data in a short time without transmission delay caused by a conventional network delay, and to send and receive data and messages regardless of network traffic. Can maintain a constant speed.

In addition, by using the RFM library to send and receive messages or data, each host can achieve the same effect as writing and reading data into its main memory.

Claims (20)

In a redundant system consisting of a plurality of hosts operating in an active or standby state and connected to each other via a network, each host is A reflective memory (RFM) having a data transmission / reception port and having a function of storing or reading data in hardware; and a data transmission / reception port of the reflective memory provided in the host. Reflective memory data transfer means connected to synchronize the contents stored in the plurality of reflective memories in the same manner in real time And data generated by a process running in an active host among the plurality of hosts are provided in the reflective memory of the standby host synchronously connected by the reflective memory of the active host and the reflective memory data transfer means. Redundancy system characterized by sharing the memory contents during active / standby switching by writing together. The redundancy system of claim 1, wherein the reflective memory is mounted in a PCI slot of the host. The redundancy system of claim 1, wherein the reflective memory data transmission means connected to the transmission / reception port of the reflective memory comprises an optical cable. The method of claim 1, wherein the reflective memory An area for storing data generated by a reflective memory device driver to operate the reflective memory; An area storing a start memory address of an area available to a user among the memory areas of the reflective memory; An area for storing data for the network communication; And An area for storing local data area information of an area where data for the network communication is stored Redundancy system, characterized in that consisting of. The method of claim 4, wherein the local data area information of the area where the data for network communication is stored A starting memory address of a local data area of an area in which data for the network communication is stored; A memory size of a local data area of an area in which data for the network communication is stored; And Separator to distinguish data for network communication Redundancy system comprising a. The redundancy system of claim 4, wherein the memory area of the reflective memory further includes an area for storing data capable of inferring the delimiter capable of distinguishing data for the network communication. In a method for performing data communication for sharing data generated by a running process with a standby host in an active host of a redundant system consisting of a plurality of hosts operating in an active or standby mode, Creating a server process or client process on the host; Allocating a local data area to a memory area of the reflective memory mounted in the host by using a reflective memory library; The client process writing a message to the allocated local data area using a reflective memory library; The server process reads a message of the client process from the allocated local data area using a reflective memory library and writes a response message using the reflective memory library; And The client process reading a response message of the server process from the allocated local data area using a reflective memory library Data communication method comprising a. The reflective memory library of claim 7, wherein the reflective memory library is used to allocate a local data area to a memory area of the reflective memory mounted to the host and to read or store data in the allocated local data area. A function of opening the reflective memory device; A function of closing the reflective memory device; A function of initializing a memory area of the reflective memory; A function for reading out a start address of the allocated local data area, having a delimiter for distinguishing data for the network communication; A function for obtaining a pointer on a main memory having a starting address of the read local data area and pointing to the read local data area; Setting a semaphore locked state of the delimiter; A function for releasing the semaphore lock state of the delimiter; And Forcibly synchronizing contents of data stored in the reflective memory mounted in the plurality of hosts; A data communication method comprising any one or a combination thereof. The method of claim 8, wherein the function of initializing a memory area of the reflective memory comprises: Mapping a memory area of the reflective memory mounted on each of the plurality of hosts to a main memory; Allocating an area for storing a start memory address of an area available to a user among the memory areas of the reflective memory; Creating a semaphore for a memory area of the reflective memory and setting an initial value; Allocating an area for storing local data area information of an area in which data for the network communication is stored; And Storing a start memory address of an area storing data for the network communication in an area storing a start memory address of an area available to a user; Data method comprising a. The method of claim 8, wherein the function of forcibly synchronizing contents of data stored in the reflective memory mounted in the plurality of hosts rewrites the contents stored in the reflective memory of any one of the plurality of hosts. And synchronizing the contents stored in the reflective memory mounted in the two hosts. The method of claim 7, wherein the allocating a local data area using a reflective memory library to a memory area of the reflective memory mounted in the plurality of hosts comprises: Initializing a memory area of a reflective memory mounted in the plurality of hosts; Generating, by the server process, a separator capable of distinguishing data generated by the server process and a client process for the network communication; The server process assigning the delimiter to the client process; And The server process and the client process allocate a local data area to the memory area of the reflective memory to store data generated by the server process and the client process for the network communication by using the delimiter. Storing information in a memory area of the reflective memory Data communication method comprising a. The data communication method of claim 11, wherein the initializing a memory area of the reflective memory mounted in the plurality of hosts uses the initialization function included in the reflective memory library. 12. The method of claim 11, wherein the step of generating, by the server process, a delimiter capable of distinguishing data generated by the server process and the client process for the network communication comprises using a selected delimiter. 12. The method of claim 11, wherein the step of generating, by the server process to distinguish data generated by the server process and the client process for the network communication, the delimiter may be allocated to the memory area of the reflective memory. And determining an integer between '0' and an integer representing the maximum number of data areas. 12. The method of claim 11, wherein the step of assigning the identifier to the client process by the server process includes assigning a different identifier to each host on which the client process is running. 12. The method of claim 11, wherein the step of assigning the separator to a client process by the server process includes assigning the same separator to all hosts on which the client process is running. 12. The method of claim 11, wherein the step of assigning the separator to a client process by the server process comprises writing the separator to a memory area of the reflective memory. 8. The method of claim 7, wherein writing or storing data or a message in a memory area of the reflective memory by using a reflective memory library Reading a start address of the allocated local data area by executing a function of reading a start address of the allocated local data area with a delimiter for the data to be written or stored; Obtaining a pointer on the main memory pointing to the read local data area by executing a function having a start address of the read local data area and obtaining a pointer on the main memory pointing to the read local data area; Setting the semaphore lock state by executing a function of setting a semaphore lock state for the delimiter for the data to be written or stored; Writing or storing a message or data in the local data area with a pointer on main memory obtained with a start address of the read local data area; And Releasing the semaphore lock state by executing a function to release a semaphore lock state for the delimiter for the data to be written or stored Data communication method comprising a. The method of claim 7, wherein the reading of the data or the message to the memory area of the reflective memory using the reflective memory library is performed. Reading a start address of the allocated local data area by executing a function of reading a start address of the allocated local data area with a delimiter for the data to be written or stored; Obtaining a pointer on the main memory pointing to the read local data area by executing a function having a start address of the read local data area and obtaining a pointer on the main memory pointing to the read local data area; Setting the semaphore lock state by executing a function of setting a semaphore lock state for the delimiter for the data to be written or stored; Reading a message or data into the local data area with a pointer on main memory obtained with a start address of the read local data area; And Releasing the semaphore lock state by executing a function to release a semaphore lock state for the delimiter for the data to be written or stored Data communication method comprising a. 8. The method of claim 7, wherein storing, writing, or reading a message or data in the allocated local data area by the server process or client process further comprises recognizing a state change of the semaphore for the delimiter. Communication method.