JP2001216282A - Server, client, client-server system, load distribution method, and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform the load distribution and fault evasion of a server in a client-server system. SOLUTION: In this client-server system, by providing a load condition monitoring part 12 for periodically monitoring the load conditions of the present server and notifying the other servers of the condition and a priority order judgment part 13 for receiving the load conditions from the respective servers and deciding the priority order of the server to respond on the respective servers, the respective servers recognize their own response priority orders and simultaneously recognize other participating servers. Also, a virtual address management part 15 for notifying the physical address of the present server as the physical address of a virtual server 4 in the case that the priority of the present server is judged in the priority order judgment part 13 in the case that a packet for requesting the physical address of the virtual server 4 flows from the client is provided and a function for turning the packet flowing from the client to the virtual server 4 to the present server is provided.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for distributing servers for which a client requests a service in a client-server system, and in particular, by changing a physical address to be notified to a client by using a virtual address as a server address. The present invention relates to a load distribution method for distributing server loads.

[0002]

2. Description of the Related Art In a conventional load distribution method in a client-server system, a plurality of servers performing the same service are connected on a network, and processing is distributed to these servers. One of the most widely used methods of distributing load is to register multiple addresses under the same host name in a name server that converts host names to addresses, and then use the name when the client converts server names to addresses. The server announces the address of a different server,
Alternatively, the method is such that the addresses of a plurality of servers are notified, and the client selects one address from the addresses to distribute the processing.

Japanese Patent Application Laid-Open No. 9-319689 discloses a "server selection method" in which a load distribution method using a name server performs load distribution or failure avoidance in consideration of server load conditions or failures.

In Japanese Patent Laid-Open Publication No. Hei 11-39267, "Load distribution method of server system", the number of used terminals in the server use terminal information of the server terminal address and the number of reference use terminals are checked, and the number of user terminals indicates the number of reference terminals. When the number of terminals exceeds the limit, a method is disclosed in which the number of terminals used by other service providing servers and the number of reference terminals are checked, and the number is assigned to the server terminal address user terminal of the service providing server with the least number of users.

[0005]

In the above-mentioned conventional server load distribution method, it is necessary to register addresses of a plurality of servers in a name server. Even when a server is added or deleted, setting of the name server is performed. Need to change. Also, the conventional method is based on the premise that an environment in which a name server can be used is a major premise. In an environment in which a name server cannot be used, there is a problem that the conventional method of specifying a server by a host name cannot be used. In such an environment, as a result, it is necessary for each client to have a different server address fixedly.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-mentioned problems, so that it is possible to easily add and delete servers without the need for centralized management for load distribution, and to provide services. An object of the present invention is to provide a system capable of distributing a load by using only one address of a server to be disclosed to a requesting client.

[0007]

A client according to the present invention is a client in a client-server system that receives a service request and a response to the service from a client to a server via a network, wherein the client is a client of the server. A virtual address having an address translation table for performing a translation to a physical address, wherein the client makes a request to the server for a physical address of the server using a virtual address of the server, receives the physical address from a specific server, The service request is registered to the address conversion table and sent to the server by the physical address.

The first server of the present invention is a server in a client-server system in which a client sends a service request and a response to the service via a network to the server, and the server periodically checks the load status of the server. A load status monitoring unit for monitoring and transmitting packets to all the servers connected to the network, and determining a priority of the server that receives and responds to the packet of the load status from the server A service request for a virtual server having a virtual address is issued by the client, and when the priority determination unit determines that the server is prioritized, the server determines the physical address of the server as a physical address of the virtual server. Virtual address management for notifying the client of a physical address Has the door, it has the function of directing the local server the service request to the virtual server from the client.

A second server of the present invention is a server in a client-server system for transmitting a service request and a response to the service from a client to the server via a network, wherein the server receives the service request from the client. A service processing unit for processing, a load status monitoring unit, a priority determining unit, a virtual address management unit, and a connection management table, wherein the load status management unit periodically updates the CPU utilization and memory of the server. Inspect the usage rate, measure the load of the server, send the load status to the network as a packet, the priority determination unit has a priority table including a plurality of entries, The load status of the load status management unit and the load of the other server from the network Means for receiving and registering in the priority order table in ascending order of the load, wherein when the physical address request of the server is issued from the same client within a predetermined time from the virtual address management unit, the priority order is Means for deleting the least loaded entry in the table; determining whether the least loaded entry has expired; and deleting the least loaded entry if the entry has expired. Confirmation means, wherein the virtual address management unit monitors a request for the physical address for a virtual server having a virtual address, and when there is a request for the physical address for the virtual server, the connection management table If the client address is registered, the server Sending an address to the client, and if the address of the client is not present in the connection management table, retrieves the physical address of the server from the least loaded entry in the priority table to the priority determination unit If it matches the physical address of its own server, it sends the physical address to the client.

A first client-server system of the present invention is a client-server system for transmitting a service request and a response to the service from a client to a server via a network, wherein the client is a client of the present invention, Is a client server system characterized by being the first server of the present invention.

A second client-server system according to the present invention is a client-server system for transmitting a service request and a response to the service from a client to a server via a network, wherein the client is the client of the present invention, Is a client-server system according to the second server of the present invention.

The load distribution method of the present invention is a load distribution method for a server in a client-server system in which a client sends a service request and a response to the service via a network from a client to the server. A first step of checking a CPU usage rate and a memory usage and measuring a load, a second step of transmitting the status of the load of the server to all the servers connected to the network, A third step of setting an expiration date of the status of the server, a fourth step of receiving the status of the load of the other server, and examining a physical address of the server that has transmitted the status of the load, and a priority table. If the same physical address exists, delete the entry, and then delete the entry in the A fifth step of adding, a sixth step of performing an adding process without performing a process of deleting the entry when the same physical address does not exist, and receiving a request for the physical address from the client for a predetermined time in the past. If there is a request within, the seventh step of deleting the entry with the smallest load from the priority table, and whether the entry with the smaller load in the priority table has expired An eighth step of checking the packet requesting the physical address of the server by the client, and a tenth step of checking whether the address of the client exists in the connection management table.
And the eleventh step of transmitting the physical address of the own server as a top priority packet so that another server does not reply, regardless of the order according to the load situation if the address exists. If the address of the client does not exist in the connection management table, a twelfth step of inquiring about the physical address of the server with the least load, and the physical address obtained in the twelfth step is A thirteenth step of checking whether the physical address is the same as the physical address of the server, and if the physical address is the same as the physical address of the own server, then the highest priority packet flows from the other server Check that the highest priority packet is not flowing, It consists fourteenth step of replying to said physical address of the server for.

The recording medium of the present invention has recorded thereon a program for causing a computer to execute each step of the load distribution method of the present invention.

[0014]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of a client server system according to an embodiment of the present invention. Referring to FIG. 1, an embodiment of the present invention provides a server A1, a server B2, and a server C3 that provide the same service.
And a client A5, a client B6, and a client C7 requesting a service are connected on the same computer network 9.

Each server processes a process in response to a request from a client and returns a response. A load status monitor that periodically checks the load status of its own server and transmits the information to the network 9. 12, a priority determination unit 13 having a priority table 131 for receiving the load status information of each server and determining the priority of each server; a connection management table 14 for managing connection information with clients; The virtual address management unit 15 returns the physical address of the server in response to a packet requesting a physical address corresponding to the address.

Each client comprises a service request unit 51 for requesting a service for the address of the virtual server 4 and an address conversion table 52 for associating the address with a physical address.

Next, the overall operation of the present embodiment will be described in detail with reference to the drawings. FIG. 2 is a flowchart showing an operation of a process performed by the priority determining unit 13 mounted on each server in the client server system according to the embodiment of the present invention. FIG. 3 is a flowchart showing an operation of processing by the virtual address management unit 15 in the load distribution method according to the embodiment of the present invention.

Referring to FIG. 1, server A1, server B
2 and the server C3 are connected on the same computer network 9. At this time, each server is connected to the network 9 at an address different from the addresses of the other servers and the virtual server 4.

On each server, the load status monitoring unit 12 periodically checks the CPU usage rate and the memory usage of the server, and determines a value representing the overall load of the server. After transmitting the acquired load status of the server as a broadcast message on the network 9, the load status monitoring unit 12 notifies the same information to the priority determination unit 13 (step 801). The contents of this notification include the address of the server, the load status, and the expiration date of this information, and the expiration date is the time until the next load status is confirmed. The load status monitoring unit 12 repeats this process.

The priority order judging unit 13 of each server receives the notification of the load status from the load status monitoring unit 12 of its own server and the notification of the load status of the other servers flowing as a broadcast message on the network 9 ( Step 802). The server address of the received load status information is checked, and if the same address exists in the priority order table 131 (step 803), the entry is deleted (step 804), and the list is arranged in the order of the load status values. Thus, the entry of the information is added (step 805). At this time, if the same address does not exist, the deletion processing is not performed, and the addition processing is performed (step 80).
Perform only 5). This makes it possible to add a server by simply connecting a server having the same function to the network 9.

When the priority determination unit 13 receives a priority inquiry from the virtual address management unit 15 (step 806), it checks whether there has been a physical address request from the same client within a predetermined time in the past (step 807). I do. If there is a request, the entry with the highest priority is deleted from the priority table 131 (step 808). This is because if there is a physical address request from the same client within a certain period of time in the past, the server with the highest priority that should return a response to the physical address has not returned a response, and the server cannot respond due to some failure It is determined that it is. As a result, the server having the second priority can respond to the next physical address request from the same client, and smooth failure avoidance can be achieved.

Next, the priority determining unit 13 checks whether the information of the entry with the highest priority in the priority table 131 has expired (step 809). If the expiration date has expired, the information has not been updated, so it is determined that the server has become unable to communicate due to some kind of failure or that the server has been removed from the network 9, and the entry from the priority table 131 is deleted. It is deleted (step 810). This makes it possible to avoid failures and at the same time facilitates removal of the server. The address of the server with the highest priority determined in this way is returned to the virtual address management unit 15 (step 81).
1) Yes.

The virtual address management unit 15 of each server
Then, each client monitors a packet requesting a physical address corresponding to the address of the virtual server 4 transmitted on the network 9 (step 901). When a physical address request corresponding to the address of the virtual server 4 flows from the client A5, first, it is checked whether or not the client address exists in the connection management table 14 (step 902). Since it is necessary to return the physical address of the own server irrespective of the order, the highest priority notification packet is transmitted to the network 9 so that the other server does not reply (step 903). This is because the address translation table 52 on the client side is provided while the service is being provided to the client.
This takes into account the case where the entry has been deleted from the server, thereby preventing a packet to the server from being transmitted to another server while the service is being provided.

If the address of the client A5 does not exist in the connection management table 14, the priority determining unit 13 is queried for the address of the server with the highest priority at the present time (step 904). It is checked whether the obtained address is the same as the address of its own server (step 905). If it is different, the response of the physical address and the provision of the service are entrusted to another server, and the virtual address management unit 15 returns to the physical address again. Waits for an address request.

If the address of the server with the highest priority is the same as the address of its own server, it is next checked whether or not the highest priority packet is flowing from another server (step 906). When the highest-priority packet flows, the response of the physical address is left to another server, and the server waits for a request for the physical address. If the highest priority packet is not flowing, the physical address of the server A1 is returned to the client A5 as the physical address of the virtual server 4 (step 907).

The client A5 receives the response of the physical address from the server A1, and registers the physical address of the server A5 in the address conversion table 52 as the physical address of the virtual server 4. This allows client A5
Is actually transmitted to the server A1, and the service processing unit 11 on the server A1 receives the request.

Each of the above steps can be realized as a program of the computer 60. FIG. 4 is a diagram showing a recording medium 61 on which the load distribution method of the present invention is recorded.

[0029]

As described above, the first effect of the present invention is that a host to which a physical address corresponding to the address of a virtual server is actually assigned a different address is transmitted instead. The presence of a plurality of servers makes it possible to perform load distribution and failure avoidance among the plurality of servers even though the service providing server notifying each client has only one address.

The second effect is that there is no server for centralized management and no special processing is required for adding / deleting servers, so that management becomes easy.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a client server system according to an embodiment of the present invention.

FIG. 2 is a diagram showing a priority determining unit 13 mounted on each server in the client server system according to the embodiment of the present invention;
Is a flowchart showing the operation of the process according to.

FIG. 3 is a flowchart showing an operation of processing by a virtual address management unit 15 in the load distribution method according to the embodiment of the present invention.

FIG. 4 is a recording medium 61 on which the load distribution method of the present invention is recorded.
FIG.

[Explanation of symbols]

 1 server A 2 server B 3 server C 4 virtual server 5 client A 6 client B 7 client C 9 network 11 service processing unit 12 load status monitoring unit 13 priority judgment unit 14 connection management table 15 virtual address management unit 51 service request unit 52 Address conversion table 60 Computer 61 Recording medium 131 Priority table

Claims (7)

[Claims] 1. A client in a client-server system for receiving a service request and a response to the service from a client to a server via a network, wherein the client is an address for converting the server into a virtual address physical address. Having a translation table, the client makes a physical address request of the server to the server using a virtual address of the server, receives the physical address from the specific server, registers the physical address in the address translation table, and registers the physical address with the physical address. A client for making the service request to the server. 2. A server in a client-server system in which a client sends a service request and a response to the service via a network to the server, wherein the server periodically monitors the load status of the server and converts the load into packets. A load status monitoring unit that transmits to all the servers connected to the network, and a priority determination unit that determines the priority of the server that receives and responds to the packet of the load status from the server, and a virtual address. When the service request for the virtual server having is issued from the client, and when the priority determination unit determines that the own server has priority, the physical address of the own server is assigned to the client as the physical address of the virtual server. A virtual address management unit for notifying,
A server having a function of directing the service request from the client to the virtual server to the server. 3. A server in a client-server system that sends a service request and a response to the service from a client to the server via a network, wherein the server is a service processing unit that processes the service request from the client. A load status monitoring unit, a priority order determination unit, a virtual address management unit, and a connection management table. The load status management unit periodically updates the CPU of the server.
Inspect the usage rate and the memory usage rate, measure the load of the server, send the load status to the network as a packet, the priority determination unit has a priority table including a plurality of entries, Means for receiving the load status of the other server from the load status management unit and the network in the own server, and registering the received load status in the priority order table in ascending order of the load; and Means for deleting the least loaded entry in the priority table when the physical address request is issued from the same client within a predetermined time, and the expiration date of the least loaded entry Is not expired, and if the entry has expired, the entry with the least load is deleted. The virtual address management unit monitors a request for the physical address for a virtual server having a virtual address, and when there is a request for the physical address for the virtual server, the connection management table If the address of the client is registered, the physical address of the server is transmitted to the client, and if the address of the client is not present in the connection management table, the priority determining unit determines the physical address in the priority table. A server that extracts the physical address of the server from the entry with the smallest load, and sends the physical address to the client if the physical address matches the physical address of the server. 4. A client-server system for transmitting a service request and a response to the service from a client to a server via a network, wherein the client is the client according to claim 1, and the server is the server. A client-server system characterized in that the server is a server. 5. A client-server system for transmitting a service request and a response to the service from a client to a server via a network, wherein the client is the client according to claim 1, and the server is the server. A client-server system characterized in that the server is a server. 6. A load distribution method for a server in a client-server system in which a client sends a service request and a response to the service via a network from a client to the server, the method comprising: A first step of checking a usage amount and measuring a load, a second step of transmitting the load status of the server to all the servers connected to the network, and an expiration date of the load status. A third step of setting, a fourth step of receiving the status of the load of the other server, and examining a physical address of the server that has transmitted the status of the load, and In some cases, after the entry is deleted, a fifth step of adding the entry in the order of the smaller load is performed. A sixth step of performing the addition processing without performing the processing of deleting the entry when the same physical address does not exist; and receiving the request for the physical address from the client and receiving the request within a predetermined time in the past. If so, a seventh step of deleting the entry with the smallest load from the priority table and an eighth step of checking whether the entry of the smaller load in the priority table has expired. A ninth step in which the client monitors a packet requesting the physical address of the server, and a first step of checking whether the address of the client exists in the connection management table.
0 and an eleventh step of transmitting the physical address of the own server as the highest priority packet so that other servers do not reply, irrespective of the order according to the load situation if the address exists. And when the address of the client does not exist in the connection management table, a twelfth step of inquiring about the physical address of the server with the least load, and the physical address obtained in the twelfth step is A thirteenth step of checking whether the physical address is the same as that of the own server, and if the physical address is the same as the physical address of the own server, Check if it is flowing and if the highest priority packet is not flowing, 14th step load balancing method of a client-server system comprised of replying the physical address of the server for bets. 7. A recording medium on which a program for causing a computer to execute the steps according to claim 6 is recorded.