JP2006287605A - Load balancer and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a load balancer which is capable of maintaining access to a server when a fault occurs. <P>SOLUTION: The load balancer comprises: a first communication means for communicating with a first network to which a plurality of servers are connected; a second communication means for communicating with a second network to which a client is connected, and which operates according to the same protocol as that of the first network; a load distribution means which selects one of the plurality of servers to supply data transmitted from the second communication means to the first communication means and transfers the data to the selected server; and a short-circuit means which short-circuits the first communication means and the second communication means and connects the first network and the second network without interposing a load distribution means. The same IP address as an IP address of one of the plurality of servers is set to the second communication means. When a fault occurs, the short-circuit means is short-circuited. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a load balancer for distributing a load between servers and clients via an IP (Internet Protocol) network.

2. Description of the Related Art Conventionally, in a server / client system, a load balancing system using a load balancer that appropriately allocates a load to each server has been used in order to alleviate the load concentration on a specific server and distribute the load. For example, see Patent Document 1).
JP 2003-174473 A

In many cases, a load distribution system using a load balancer includes a backup load balancer 101 in addition to the main load balancer 100, as shown in FIG. Then, when a failure (failure, hang-up, increase in data amount, power failure, etc.) occurs in the main load balancer 100, a response of switching to the backup load balancer 101 is performed.
If there is no backup, it is necessary to take measures such as changing the IP addresses of the web servers 200 and 201.
As described above, in the conventional load distribution system, the handling at the time of the occurrence of the trouble is complicated. In addition, when the response is delayed, access to the web server may become impossible.

  The present invention has been made in view of the above circumstances, and an object thereof is to provide a load balancer and a program that can maintain access to a server even when a failure occurs.

  The load balancer according to the present invention includes a first communication unit that communicates with a first network to which a plurality of servers are connected, and a client that is connected and operates according to the same protocol as the first network. Second communication means for communicating with the second network, and data to be transmitted from the second communication means to the first communication means in accordance with load amounts of the plurality of servers. A load balancer configured to select which of them to supply and transfer data to the selected server, wherein the first communication unit and the second communication unit And short circuit means for connecting the first network and the second network without going through the load distribution means, and the second communication means includes a plurality of front terminals. Any one of the server's IP (Internet Protocol) address and the same IP address of the server is set, characterized in that.

  And further comprising a short-circuit control means for supplying instruction information to instruct the short-circuit means to short-circuit the first communication means and the second communication means when a failure occurs in the load balancer. In response to the information, the short-circuit means may short-circuit the first communication means and the second communication means.

  The short-circuit control unit may supply the instruction information to the short-circuit unit when there is no response from the load distribution unit even after a predetermined time has elapsed.

  The short-circuit control unit may supply the instruction information to the short-circuit unit when power is not supplied to the load balancer.

  The load balancer receives an address resolution request for obtaining a MAC address from the IP address in a state where the server in which the same IP address as that of the second communication unit is set is operating normally. Storing the address resolution request and the address resolution response transmitted by the server in response to the address resolution request and storing the address resolution response as address resolution information; and storing the address resolution request in the holding unit when a failure occurs in the server And proxy response means for transmitting an address resolution response on behalf of the server based on the address resolution information.

A program according to a second aspect of the present invention is the same as the first network, in which a computer is connected to a first communication means for communicating with a first network to which a plurality of servers are connected, and a client is connected. The second communication means that communicates with the second network that operates according to the protocol of the data, and data that is transmitted from the second communication means to the first communication means according to the load amount of the plurality of servers To which of the plurality of servers is selected, and the load distribution means for transferring data to the selected server, the first communication means, and the second communication means are short-circuited A program that causes a load balancer to function as a load balancer comprising: a short-circuit unit that connects the first network and the second network without going through the load distribution unit; Wherein the second communication means, any one of the server's IP (Internet Protocol) address and the same IP address of the plurality of the server is set, characterized in that.

  The load balancer and program according to the present invention short-circuit a network to which a server is connected and a network to which a client is connected. Therefore, the load balancer and program according to the present invention can maintain access to the server even when a failure occurs.

  An example in which the load balancer of the present invention is applied to the load distribution system 1 will be described with reference to the drawings.

  As shown in FIG. 1, the load distribution system 1 includes a load balancer 100, a web server 200, a web server 201, a client 300, a firewall 400, a communication network A, and a communication network B. The

  A load balancer 100, a web server 200, and a web server 201 are connected to the communication network A.

  The communication network B is a LAN (Local Area Network) such as a so-called intranet. A load balancer 100, a client 300, and a firewall 400 are connected to the communication network B.

  Communication network A and communication network B are communication networks of the same protocol.

  The firewall 400 is a network connection management device that manages mutual access between the communication network B and an external network.

  The client 300 is a terminal used by a user who receives a service supplied from the web servers 200 and 201 and is connected to the communication network B.

  The web server 200 and the web server 201 are web servers that provide the same service to the client 300 via the network, and are connected to the communication network A. At this time, the IP address in the communication network A of the web server 200 is set to X, and the IP address in the communication network A of the web server 201 is set to Y.

  As shown in FIG. 2, the load balancer 100 includes a control unit 110, a storage unit 120, an input unit 130, an output unit 140, and a communication control unit 150.

The input unit 130 includes, for example, a keyboard, a mouse, and the like, receives an operation by a user, and supplies operation information to the control unit 110.
The output unit 140 includes, for example, a display device, a printer, and the like, and provides output information to the user under the control of the control unit 110.

  As shown in FIG. 3, the communication control unit 150 includes a LAN port 154A, a LAN port 154B, a relay unit 151A, a relay unit 151B, a communication circuit 152A, a communication circuit 152B, and a short-circuit relay 153. Is done.

The LAN port 154A is connected to the communication network A, and the LAN port 154B is connected to the communication network B.
At this time, the IP (Internet Protocol) address in the communication network A of the LAN port 154A is set to Z. The IP address in the communication network B of the LAN port 154B is set to X.

As described above, in the load distribution system 1 of the present embodiment, the IP address in the communication network B of the LAN port 154B of the load balancer 100 and the IP address in the communication network A of the web server 200 are set to be the same. The
The IP address X in the communication network B of the LAN port 154B is a virtual IP address when the client 300 accesses the web servers 200 and 201.

The relay unit 151A is a switch that selectively connects either the contact A or the contact C to the contact B. Which contact is selected is selected based on the control of the control unit 110.
The relay unit 151B is a switch that selectively connects either the contact point a or the contact point c with the contact point b, similarly to the relay unit 151A. Which contact is selected is selected based on the control of the control unit 110.

  The communication circuit 152A controls communication via the communication network A of the load balancer 100 when the relay unit 151A connects the contact A and the contact B. More specifically, the communication circuit 152A receives all the packets from the web server 200 or 201 on the communication network A and supplies them to the control unit 110. The packet supplied from the communication circuit 152A and the control unit 110 is sent to the communication network A.

  The communication circuit 152B controls communication via the communication network B of the load balancer 100 when the relay unit 151B connects the contact point a and the contact point b. More specifically, when receiving a packet whose destination IP address is X from the client 300 via the communication network B, the communication circuit 152B supplies this to the control unit 110. In addition, the communication circuit 152B sends the packet supplied from the control unit 110 to the communication network B.

The short-circuit relay 153 is a changeover switch that can connect / disconnect between the contact X and the contact Y by manual operation. As shown in FIG. 3, the contact X is connected to the contact C of the relay unit 151A, and the contact Y is connected to the contact c of the relay unit 151B.
The short-circuit relay 153 enables the data through function described later by connecting the contact X and the contact Y, and disables the data through function by setting the disconnection state.

The control unit 110 includes a CPU (Central Processing Unit), a timer for measuring a predetermined time, and the like, and controls the overall operation of the load balancer 100 according to a program stored in the storage unit 120.
Specifically, the control unit 110 controls the entire load balancer 100 so as to realize, for example, a load balance function, a data through function, and an ARP (Address Resolution Protocol) proxy response function described below.

  The control unit 110 transfers the packet whose destination IP address is X supplied from the communication circuit 152B to any web server so that the load is not concentrated on one side according to the load amount of the web servers 200 and 201. Realize a load balance function to select whether to do.

  The control unit 110 realizes a data through function by controlling the relay unit 151A and the relay unit 151B in accordance with the presence or absence of a CPU response within a predetermined period.

  If a failure occurs in the web server 200 and an ARP request for a MAC (Media Access Control) address corresponding to the IP address X is generated, the control unit 110 stores the ARP information stored in the storage unit 120 in advance. Based on the above, an ARP proxy response function for transmitting an ARP response on behalf of the web server 200 is realized.

  In the load distribution system 1 configured as described above, the load balance function, the data through function, and the ARP proxy response function realized by the load balancer 100 will be described in detail.

  In order to realize the load balance function, the control unit 110 of the load balancer 100 operates as follows.

  That is, when the control unit 110 receives a packet whose destination IP address is X, the control unit 110 acquires the load amount of the web server 200 and the load amount of the web server 201 via the communication circuit 152A and the communication network A. Then, the control unit 110 selects a web server with a small load as a packet transfer destination.

When the web server 200 is selected as the packet transfer destination, the control unit 110 supplies the packet to the communication circuit 152A as it is.
In addition, when the web server 201 is selected as a packet transfer destination, the control unit 110 changes the destination IP address of the packet to Y, and then supplies the changed packet to the communication circuit 152A.

  It should be noted that the controller 110 operates as follows when communicating in the opposite direction to when the load balance function is executed, that is, when sending a packet from the communication network A to the communication network B. That is, when the control unit 110 receives a packet sent from the web server 200 from the communication circuit 152A, the control unit 110 sends the packet to the communication network B as it is via the communication circuit 152B. When the packet sent by the web server 201 is received, the source IP address of the packet is changed to X, and then the changed packet is supplied to the communication circuit 152B.

  In order to implement the data through function, the control unit 110 of the load balancer 100 operates as follows.

  That is, the control unit 110 monitors whether or not there is a response from the CPU within a predetermined period measured by a timer. Then, when there is a CPU response within a predetermined period, the control unit 110 connects the contact A and the contact B of the relay unit 151A and the contact a and the contact b of the relay unit 151B, respectively. In this case, a packet transmitted between the communication network A and the communication network B passes through the control unit 110. Therefore, a packet transmitted from the communication network B to the communication network A is transmitted to either the web server 200 or 201 by the above-described load balance function (data through function does not work).

  On the other hand, when the CPU does not respond within a predetermined period, control unit 110 connects contact C and contact B of relay unit 151A and contact c and contact b of relay unit 151B. In this case, a packet transmitted between the communication network A and the communication network B passes through the short circuit relay 153. Therefore, the load balance function described above is not used, and the destination IP address of the packet is always X. That is, all packets are transmitted to the web server 200 (data through function works).

  When the short-circuit relay 153 is in a disconnected state, even if the contact C and contact B of the relay unit 151A and the contact c and contact b of the relay unit 151B are connected, the packet is communicated with the communication network A. Transmission to and from network B is not possible. That is, the data through function is disabled.

  In order to realize the ARP proxy response function, the control unit 110 of the load balancer 100 operates as follows.

  First, prior to executing the ARP proxy response function, when the web server 200 is functioning normally, the control unit 110 executes the normal response process as shown in the flowchart of FIG. The MAC address of the server 200 is acquired.

When a request for requesting a MAC address corresponding to the IP address X (ARP request) is generated from the client 300 while the web server 200 is functioning normally, the load distribution system 1 starts normal response processing.
When the normal response process is started, the control unit 110 stores the ARP request in the storage unit 120 (step S100). Then, the control unit 110 transfers the ARP request to the communication network A via the communication control unit 150 without changing the content of the ARP request (step S110).

  Then, when the web server 200 transmits the result (ARP response) to the ARP request (step S120), the control unit 110 receives this via the communication control unit 150, and determines the MAC address included in the ARP response as follows. The data is stored in association with the previously stored ARP request (step S130). Also, the control unit 110 transfers the ARP response to the communication network B to the requesting client 300 (step S140). The client receives the ARP response (step S150), and the normal response process ends.

  The control unit 110 monitors whether the web server 200 is functioning normally. And when it detects that the web server 200 is not functioning normally, the control part 110 implement | achieves an ARP proxy response function by performing the proxy response process shown in FIG.

When the client 300 generates an ARP request for the IP address X, the load distribution system 1 starts proxy response processing.
In response to the ARP request, the control unit 110 sends an ARP response on behalf of the proxy using the MAC address acquired in advance when the web server 200 is operating normally (step S200).

  When the client 300 receives the ARP response (step S210), the client 300 generates a connection request (step S220). Upon receiving the connection request, the control unit 110 generates an ARP request for the web server 201 via the communication control unit 150 (step S230). In response to this, the web server 201 generates an ARP response (step S240). The control unit 110 receives the ARP response (step S250) and transmits a connection request to the web server 201 (step S260). In response to this, the web server 201 establishes a connection with the client 300 (step S270). This completes the proxy response process.

  When the web server 200 is restored by replacing it with another hardware, the control unit 110 of the load balancer 100 executes normal response processing and updates the ARP information.

  Next, with reference to the flowchart shown in FIG. 6, the operation at the time of failure of the load distribution system 1 performed when the data amount supplied to the load balancer 100 increases and the data through function operates will be described. I will explain it.

  When the data through function is activated, the communication network A and the communication network B are physically short-circuited via the relay unit 151A, the short-circuit relay 153, and the relay unit 151B in the communication control unit 150 of the load balancer 100. Has been.

  In this state, when the client 300 generates an ARP request for X, which is the virtual IP address of the web servers 200 and 201 (that is, the IP address of the LAN port 154B of the load balancer), the operation when a failure occurs Is started. The ARP request is received by the LAN port 154B of the load balancer 100, but is directly sent from the LAN port 154A to the communication network A via the relay unit 151B, the short-circuit relay 153, and the relay unit 151A (step S300). .

  The ARP request sent to the communication network A is received by the web server 200 because the destination IP address is X, and the web server 200 generates an ARP response (step S310).

  The ARP response is received by the LAN port 154A of the load balancer 100, but is directly sent from the LAN port 154B to the communication network B via the relay unit 151A, the short-circuit relay 153, and the relay unit 151B (step S320). ).

  The ARP response sent to the communication network B is received by the client 300 (step S330). Next, the client 300 generates a connection request (step S340).

  The connection request is received by the LAN port 154B of the load balancer 100, but is directly sent from the LAN port 154A to the communication network A via the relay unit 151B, the short-circuit relay 153, and the relay unit 151A (step S350). .

  The connection request sent to the communication network A is received by the web server 200, and the web server 200 establishes a connection with the client 300 (step S360). This is the end of the operation when a failure occurs.

  Thus, in the load distribution system 1 of the present embodiment, the IP address of the load balancer 100 in the communication network B and the IP address of the web server 200 in the communication network A are set to be the same. When a failure occurs, the load balancer 100 short-circuits the communication network A and the communication network B. For this reason, when the client 300 tries to access the virtual IP addresses (that is, the load balancer 100) of the web servers 200 and 201 in the short-circuit state, the access is made by the communication network A and the communication network. It is possible to convert the access to the web server 200 on the communication network A without changing the setting of B.

  As described above, the load distribution system 1 according to the present embodiment can maintain access to the web server even when a failure occurs such that the CPU hangs up and the load balancer 100 does not respond.

In the above embodiment, the case where the relay unit 151A and the relay unit 151B are switched according to whether the CPU of the load balancer 100 is functioning normally has been described as an example. However, the load balancer of the present invention may monitor the amount of data supplied to the load balancer and switch between the relay unit 151A and the relay unit 151B depending on whether the data amount exceeds a predetermined threshold. .
This makes it possible to maintain access to the web server while preventing the load balancer 100 from hanging up.

In addition, the load balancer of the present invention may switch between the relay unit 151A and the relay unit 151B in accordance with ON / OFF of the power supply of the load balancer 100.
Thereby, even when a power failure occurs in the load balancer 100, it is possible to maintain access to the web server.

  The load balancer of the present invention may include means for manually switching between the relay unit 151A and the relay unit 151B.

  In the above embodiment, the case where the load distribution system 1 includes two web servers has been described as an example. However, the number of web servers may be three or more.

  Although the embodiment of the present invention has been described above, the load balancer according to the present invention can be realized using a normal computer system, not a dedicated system. For example, to cause a computer connected to the communication network A and the communication network B to execute the operations of the control unit 110, the storage unit 120, the input unit 130, the output unit 140, and the communication control unit 150 described above. By installing these programs from a medium (CD-ROM, MO, etc.) storing these programs, the computer functions as a load balancer that executes the above-described processing.

  Further, the method for providing the program to the computer is arbitrary. For example, the program may be uploaded to a bulletin board (BBS) on a communication line and distributed to a computer via the communication line. The program may be transmitted by a modulated wave obtained by modulating a carrier wave with a signal representing the program, and a device that receives the modulated wave may demodulate the modulated wave to restore the program. Then, the computer activates this program and executes it in the same manner as other application programs under the control of the OS. Thus, the computer functions as a load balancer that executes the above-described processing.

  When the OS shares a part of the process, or when the OS constitutes a part of one component of the present invention, the recording medium may store a program excluding that part. .

It is a block diagram which shows the structure of a load distribution system. It is a block diagram which shows the structure of a load balancer. It is a block diagram which shows the structure of a communication control part. It is a flowchart which shows the procedure of a normal response process. It is a flowchart which shows the procedure of a proxy response process. It is a flowchart explaining the procedure of the operation | movement at the time of the failure occurrence of a load distribution system. It is a block diagram which shows the structure of the conventional load distribution system.

Explanation of symbols

100 Load balancer 200 Web server 201 Web server 300 Client 400 Firewall

Claims (6)

First communication means for communicating with a first network to which a plurality of servers are connected;
A second communication means for communicating with a second network, to which a client is connected and which operates according to the same protocol as the first network;
According to load amounts of the plurality of servers, a server serving as a destination to be transmitted from the second communication unit to the first communication unit is selected from the plurality of servers, and data is transmitted to the selected server. Load balancing means to transfer,
A load balancer comprising:
Short-circuit means for short-circuiting the first communication means and the second communication means, and connecting the first network and the second network without going through the load distribution means,
In the second communication means, the same IP address as the IP (Internet Protocol) address of any one of the servers is set.
A load balancer characterized by that. Short-circuit control means for supplying instruction information instructing the short-circuit means to short-circuit the first communication means and the second communication means when a failure occurs in the load balancer;
In response to the instruction information, the short-circuit means short-circuits the first communication means and the second communication means,
The load balancer according to claim 1. The short-circuit control means supplies the instruction information to the short-circuit means when there is no response from the load distribution means even after a predetermined time has elapsed;
The load balancer according to claim 2. When the power is not supplied to the load balancer, the short circuit control means supplies the instruction information to the short circuit means,
The load balancer according to claim 2 or 3, wherein When an address resolution request for obtaining a MAC address from the IP address is generated in a state where the server set with the same IP address as the second communication unit is operating normally, the address resolution is performed. A holding means for acquiring a request and an address resolution response transmitted by the server in response to the request and holding it as address resolution information;
Proxy response means for transmitting an address resolution response on behalf of the server based on the address resolution information stored in the holding means when a failure occurs in the server;
The load balancer according to any one of claims 1 to 4, wherein Computer
First communication means for communicating with a first network to which a plurality of servers are connected;
A second communication means for communicating with a second network, to which a client is connected and which operates according to the same protocol as the first network;
Select which of the plurality of servers supplies data to be transmitted from the second communication means to the first communication means according to the load amount of the plurality of servers, and the selected Load balancing means for transferring data to the server;
Short-circuit means for short-circuiting the first communication means and the second communication means, and connecting the first network and the second network without going through the load distribution means;
A program that functions as a load balancer with
In the second communication means, the same IP address as the IP (Internet Protocol) address of any one of the servers is set.
A program characterized by that.

Images