KR100989490B1 - Sysem and its method for load distribution between servers - Google Patents

Abstract

1. TECHNICAL FIELD OF THE INVENTION

The present invention relates to a load-balancing system between servers and a method thereof and a computer-readable recording medium having recorded thereon a program for realizing the method.

2. The technical problem to be solved by the invention

The present invention is a method for distributing load in processing a transaction occurring between a server and a server, by using server state information and a load balancing algorithm that is updated in real time at each server when a service request is made from one server to another server, By selecting one server from a server farm that provides the same service, a load-balancing system among the servers and a method for realizing the method and a program for realizing the method can be read by a computer that appropriately distributes the load generated between the servers. Its purpose is to provide a record carrier.

3. Summary of the Solution of the Invention

According to the present invention, in a load balancing system between servers, server farms, which are server groups in which a plurality of servers provide the same service, are hierarchically connected, and server state information updated in real time by each server when a service request is performed between upper and lower servers. And performing a service by selecting one server in the server farm by using a load balancing algorithm, each server comprising: server state management means for managing a state information table storing states of respective servers of upper and lower hierarchies; Server state information transmitting means for periodically transmitting its resource state and service provision state to respective servers of the upper and lower layers; Selecting a service performing server according to the load balancing algorithm set for the service after extracting information on the servers performing the service according to the service request from each server in the upper and lower layers using the state information table. Server selection means for; And service requesting means for requesting a service from the service performing server selected by the server selecting means.

4. Important uses of the invention

The invention is used in load balancing systems and the like.

Server farm, load balancing between each server, state information table, load balancing algorithm

Description

Load balancing system between each server and its method {SYSEM AND ITS METHOD FOR LOAD DISTRIBUTION BETWEEN SERVERS}             

1 is a configuration diagram of a server system using a server load balancing switch according to a conventional method.

2 is a configuration diagram of a server system using clustering according to a conventional method.

Figure 3 is an embodiment configuration of a load balancing system between each server according to the present invention.

Figure 4 is a detailed configuration diagram of one embodiment of each server according to the present invention.

5 is a structural diagram of an embodiment of a state information table according to the present invention;

6 is a flowchart illustrating an embodiment of a method for load balancing between servers according to the present invention.

* Explanation of symbols on the main parts of the drawing

31: Web Server Farm 32: eAI Server Farm

33: SCM Server Farm 34: ERP Server Farm

35: client 41: server state management unit

42: server status information transmission unit 43: server selection unit                 

44: service request unit

The present invention relates to a load-balancing system between each server, and a method and a computer-readable recording medium recording a program for realizing the method, which are updated in real time when a service is requested from one server to another server. A load balancing system and a method between the servers and the method for appropriately distributing the load generated between the servers by selecting one server in a server farm providing the same service by using the server state information and load balancing algorithm The present invention relates to a computer-readable recording medium having recorded thereon a program for realizing this.

Traffic jams are not only happening on roads in the real world, but traffic jams are becoming common on the Internet, which serves as an important infrastructure for our society, and the damage to each user and the cost burden of our society are not so great. .

For example, the newspaper reports that many examinees and parents rolled their feet on the message "Can't connect to the server down" on the deadline. In addition, they often suffer from system failures during online stock trading. I received my elaborate resume online, but I found it late and found it unusable due to the company's server down. As such, the number of online processing such as Internet banking and cyber trading is increasing rapidly. Therefore, the necessity of efficient system management is more urgent for the enterprise.

Server downtime, on the other hand, occurs mainly when the traffic that needs to be handled due to congestion increases. In particular, if data is gathered around a particular server or firewall through which all traffic passes, it is prone to failure. In other words, since a bottleneck occurs when excessive transactions occur in the server, a technique for properly balancing the load on the server is required.

In general, a technique for load balancing a server is to use a load balancing switch that selects a server according to a specific algorithm to distribute transactions generated from a client, and to provide redundancy and load balancing through clustering. . This will be described in detail with reference to FIGS. 1 and 2, respectively.

First, as shown in FIG. 2, a server system using clustering uses a unique algorithm between servers when a user connects to this representative IP by installing a virtual representative IP address after installing the same application program on two or more servers. The concept is to increase the performance by balancing the load on the server. In particular, this approach allows service to continue uninterrupted even in the event of a specific server failure. However, this method requires dedicated communication devices that connect shared storage (SAN, SCSI) and server-to-server on servers, and requires the installation of additional expensive load balancing application programs. There was a problem such as poor server performance. In addition, it requires difficult hardware and software configuration and operation experience, and the system becomes unstable as the number of servers bound by clustering increases.

In addition, the server system using the server load balancing switch is as shown in FIG. Commonly used as a load balancing switch is a layer 4 switch. Usually called L4 switch, it is located at the front of the server and has a public IP address (VIP) that represents the servers and requests connection to this address when a server connection request is made from outside. At this time, L4 switch plays a role of handing session to appropriate server according to requested service port. This method is still the most widely used method for server load balancing, allowing service providers to provide high performance and stable service using multiple low cost servers rather than expensive high performance servers. Recently, it is also called L4-L7 switch because load balancing service using all areas not only layer 4 but also layer 7 is possible. However, this method is expensive to switch and network change.

On the other hand, the general server load balancing methods described above are distributed only for the load on the transactions transferred from the client to the server using the Hypertext Transfer Protocol (HTTP) and the File Transfer Protocol (FTP). There was a problem that the load balancing function for the DBMS (Database Management System), the socket, the remote procedure call (RPC), etc., transmitted from the server to the server was not provided.                         

In addition, conventional server load balancing methods only support simple load balancing algorithms (Round Robin, Hash, Least Connection Scheduling, etc.), so that a flexible load balancing algorithm according to application characteristics cannot be applied. There was a problem that the dispersion is not achieved.

The present invention, which is proposed to solve the above problems, as a method for distributing the load during the transaction processing between the server and the server, updating in real time at each server when a service request from one server to another server A load balancing system and a method among the servers, and a method thereof, for properly distributing loads among servers by selecting one server in a server farm providing the same service by using the server state information and the load balancing algorithm. An object of the present invention is to provide a computer-readable recording medium having recorded thereon a program for realizing this.

The present invention for achieving the above object is, in the load balancing system between each server, a server farm, which is a server group in which a plurality of servers provide the same service is connected hierarchically, each server at the time of service request between the upper and lower servers The server status information and the load balancing algorithm are updated in real time to select a server in the server farm to perform a service, wherein each server stores a state information table that stores the state of each server in the upper and lower hierarchies. Server state management means for managing; Server state information transmitting means for periodically transmitting its resource state and service provision state to respective servers of the upper and lower layers; Selecting a service performing server according to the load balancing algorithm set for the service after extracting information on the servers performing the service according to the service request from each server in the upper and lower layers using the state information table. Server selection means for; And service requesting means for requesting a service from the service performing server selected by the server selecting means.

In addition, the present invention, in the load balancing method between each server, each server included in the server farm consisting of a plurality of servers providing the same service periodically transmits its status information to each server in the upper and lower tiers, A first step of storing / managing state information transmitted from respective servers of the upper and lower hierarchies in a state information table; A second step of extracting information on servers performing the service from a state information table by a server receiving a service; A third step of selecting, by the server, a service performing server according to a load balancing algorithm set for the service by using the extracted information of the servers; And a fourth step of the server transmitting a service request signal to the selected service performing server.

On the other hand, the present invention, each server in the server farm consisting of a plurality of servers providing the same service to each server in the load balancing system between each server periodically transmits its status information to each server in the upper and lower tiers A first function of storing / managing state information transmitted from respective servers of the upper and lower hierarchies in a state information table in a state information table; A second function of extracting information on servers performing the service from a state information table by a server receiving a service; A third function of selecting, by the server, a service performing server according to a load balancing algorithm set for the service by using the extracted information of the servers; And a computer readable recording medium having recorded thereon a program for realizing a fourth function of the server transmitting a service request signal to the selected service performing server.

The above-mentioned objects, features and advantages will become more apparent from the following detailed description in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a diagram illustrating an embodiment of a load balancing system between servers according to the present invention.

As shown in FIG. 3, each server-to-server load balancing system according to the present invention includes a web server farm 31, an enterprise application integration (EAI) server farm 32, and a supply chain management (SCM) server farm. (33), an enterprise resource planning (ERP) server farm 34. At this time, the server farm refers to a group of servers providing the same service.

First, the web server farm 31 includes a plurality of servers 311, 312, and 313 for providing various services in response to a request of a client connected through the Internet.

In addition, the enterprise application integration (EAI) server farm 32 includes a plurality of servers 321 and 322 for performing a function of integrating various applications in the enterprise based on business logic using middleware. do. Where traditional middleware has applied individual applications point-to-point, Enterprise Application Integration (EAI) is a solution that integrates, manages, and uses the necessary information by organically linking all interrelated applications in the enterprise (SCM, CRM, ERP). to be.

In addition, the supply chain management (SCM) server farm 33 establishes all schedules for the purchase, manufacture, and sales of parts, systematically organizes the customer's demand planning and logistics, and smoothes the flow of products. And a plurality of servers 331 and 332 for performing efficiently. Recently, most companies have reduced their total cost and time through supply chain management (SCM), increasing sales, flexibility and customer responsiveness.

Finally, the Enterprise Resource Planning (ERP) server farm 34 provides a number of servers to perform pure management and management support functions, including design, finance, accounting, sales, and human resources, as well as production and production management. (341, 342, 343).

Meanwhile, the client 35 requests various services by accessing one of the servers included in the web server farm 31. Then, a transaction occurs between each server farm 31 to 34 in order to provide the service requested by the client 35. At this time, in the present invention, in effectively distributing transactions occurring between the server farms 31 to 34, the software is implemented without the help of a separate hardware (load distribution switch) or software. A more detailed configuration will be described with reference to FIG. 4.

4 is a detailed configuration diagram of one embodiment of each server according to the present invention.

As shown in FIG. 4, each server according to the present invention includes a server state manager 41 for managing a state information table that stores states of upper and lower hierarchical servers, and a resource state (CPU, memory) of its own. Server status information transmission unit 42 for periodically transmitting the server share) and the service providing status to the servers of the upper and lower layers, and performing the corresponding service from the servers of the upper and lower layers or from the client. A server selector 43 for selecting a service performing server according to a load balancing algorithm set for the corresponding service after extracting information about the servers using the state information table of the server state manager 41, and server selection A service request unit 44 for requesting a service from the service performing server selected in the unit 43 is included.

The state information table stored in the server state storage unit 41 will be described later with reference to FIG. 5.

5 is a structural diagram of an embodiment of a state information table according to the present invention.

As shown in Fig. 5, the state information table according to the present invention includes a service name field 51 for uniquely distinguishing each server farm, and is assigned to each server farm for use when the service request unit 44 connects. A virtual IP field 52 for storing an address for the server, a load balancing algorithm field 53 for setting a load balancing algorithm suitable for each server farm, a server IP field 54 for storing a server's actual address, and each server A service status field 55 for storing a service provision state of the server, and a server status field 56 for storing a resource state of the server.

At this time, each detailed field of the service status field 55 and the server status field 56 is composed of different information for each service. That is, the service status field 55 of the enterprise application integration (EAI) service includes a response time 551, a running instance 552, and a sleeping instance 553. The Service Status field 55 of the Supply Chain Management (SCM) service includes an Active Session (554), an Aborted Tx (555), and Enterprise Resource Planning (ERP). The service status field 55 of the service includes a working set peak 556 and a job queued 557.

Each detailed field of the server status field 56 is also composed of different information (CPU occupancy, memory availability, etc.) according to the service, similarly to the service status field 55.

In the load balancing algorithm field 53, various load balancing algorithms are set with priority. First, if one server cannot be selected by executing the highest priority load balancing algorithm, the next one is selected by sequentially performing the load balancing algorithm with the highest priority.

Load balancing algorithms can include Round Robin, Weighted Round Robin, Hash, Least Connection Scheduling, Response Time Fast, etc. The condition may be set according to the details of the details field included in the service status field 55 and the details field included in the server status field 56.

For example, the load balancing algorithm field 55 of an enterprise application integration (EAI) service may include the service status field 55 of each server 321, 322 included in the enterprise application integration (EAI) server farm 32. Using the subfields, first select a server as Response Time First 531, if not, select a server with a running instance low 532, and fail to select If so, the server having the minimum sleeping instance low 533 is selected.

The selected enterprise application integration (EAI) 321 then first uses the subfields of the service status field 55 of each server 331,332 included in the supply chain management (SCM) server farm 33. Swapping (Pages / Sec) (534) selects a server less than 5000; if not, selects the server with the least number of unsuccessful transfers (Arorted tx); otherwise, minimum CPU Usage Select with a server (536).

In addition, enterprise application integration (EAI) 321 is first wicked, using the subfields of the service status field 55 of each server 341, 342, 343 included in the enterprise resource planning (ERP) server farm 34 again. If the working set peak (537) is less than or equal to 2000 and Disk Write Busy (537) selects the smallest server and fails to select it, the queued job (538) is greater than 100. Select a small server, and if not, select the server with the minimum network interface card utilization (539).

6 is a flowchart illustrating an embodiment of a method for load balancing between servers according to the present invention.

First of all, the load balancing program of each server manages a state information table updated at regular intervals, and periodically transmits its own state information to each server of the upper and lower layers.

First, the load balancer receives the service request signal (601), extracts information on the servers performing the service from the state information table (602), and then uses the load balancing algorithm set for the service in real time. The service performing server is selected (603). In operation 604, the service is requested to the selected service execution server and the result is returned.

The method of the present invention as described above may be implemented as a program and stored in a computer-readable recording medium (CD-ROM, RAM, ROM, floppy disk, hard disk, magneto-optical disk, etc.). Since this process can be easily implemented by those skilled in the art will not be described in more detail.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains. It is not limited by the drawings.

As described above, the present invention provides a load balancing function for transactions such as a database management system (DBMS), a socket, a remote procedure call (RPC), and the like, which are transmitted from a server to a server. As a result, load balancing between servers can be more effective.

Claims (9)

In each server load balancing system, Server farms, which are server groups in which multiple servers provide the same service, are connected hierarchically, and each server selects one of the server farms and performs a service when requesting service between upper and lower servers. Each server, Server state management means for managing a state information table storing states of respective servers of the upper and lower hierarchies; Server state information transmitting means for periodically transmitting its resource state and service provision state to respective servers of the upper and lower layers; In accordance with the service request from each server in the upper and lower layers, information on the servers performing the service is extracted by using the state information table, and then the service is performed according to the priority among a plurality of load balancing algorithms set for the service. Server selecting means for selecting a performing server; And Service requesting means for requesting a service from the service performing server selected by the server selecting means Load balancing system between each server comprising a. The method of claim 1, The state information table, A service name field for storing a service name provided by each server farm; A virtual IP field for storing a virtual IP assigned to each server farm; A load balancing algorithm field in which the plurality of load balancing algorithms are set for each server farm; A server IP field in which an actual address of each server is stored; A service status field for storing a service providing status of each server; And Server status field for storing resource status of each server Load balancing system between each server comprising a. The method of claim 2, The service status field and the server status field, Load balancing system between each server, characterized in that it comprises a different sub-field for each server farm. The method according to claim 2 or 3, The load balancing algorithm field is And said plurality of load balancing algorithms are set with priority. The method of claim 4, wherein The load balancing algorithm field is And a server selection condition is set according to a subfield included in the service status field and a subfield included in a server status field. In the load balancing method between each server, Each server included in a server farm composed of multiple servers providing the same service periodically transmits its own state information to each server in the upper and lower tiers, and receives the state information transmitted at regular intervals from the servers in the upper and lower tiers. Storing / managing the status information table; Extracting, by the server receiving the service, information about servers performing the service from a state information table; Selecting, by the server, a service performing server according to a priority among a plurality of load balancing algorithms set for the service by using the extracted information of the servers; And The server transmitting a service request signal to the selected service performing server Load balancing between each server comprising a. The method of claim 6, The state information table, A service name field for storing a service name provided by each server farm; A virtual IP field for storing a virtual IP assigned to each server farm; A load balancing algorithm field in which the plurality of load balancing algorithms are set for each server farm; A server IP field in which an actual address of each server is stored; A service status field for storing a service providing status of each server; And Server status field for storing resource status of each server Load balancing between each server comprising a. The method of claim 7, wherein The load balancing algorithm field is And the plurality of load balancing algorithms are set with priority. On each server in a load balancing system between each server, Each server included in a server farm consisting of a plurality of servers providing the same service periodically transmits its own state information to each server in the upper and lower tiers, and the state information transmitted at regular intervals from each server in the upper and lower tiers. Storing / managing a status information table; A function of extracting information on servers performing the service from a state information table by a server receiving a service; Selecting, by the server, a service performing server according to a priority among a plurality of load balancing algorithms set for the service by using the extracted information of the servers; And The server sends a service request signal to the selected service performing server A computer-readable recording medium having recorded thereon a program for realizing this.

Images