JP4515354B2 - Load balancing processing system and apparatus - Google Patents

Description

  The present invention relates to a load distribution technique for easily grasping the load status of the entire system in a client server system.

Various methods of load distribution in a client server system have been proposed (for example, Patent Document 1).
JP-A-10-275126

  However, with the conventional load distribution technique, it has become difficult to easily grasp how much load the entire system is in as a result of distributing the load. For example, when the load is distributed to a plurality of servers by round robin or the like, the load state of the entire system cannot be grasped unless the amount of load applied to all the servers is grasped.

  Accordingly, an object of the present invention is to provide a load distribution technique for easily grasping the state of a load applied to the entire system.

  A load distribution processing system according to an embodiment of the present invention includes a load distribution apparatus and a plurality of servers. The load balancer stores a priority for each server indicating a priority for assigning a process to each server, a maximum processing amount in each server, and a processing amount in execution in each server. And upon receiving the request, referring to the storage means, a distribution means for allocating the request to a server having the highest priority among servers in which the processing amount during execution does not reach the maximum processing amount, Prepare.

  In a preferred embodiment, each of the plurality of servers can execute a plurality of processes. The storage means stores the priority, the maximum processing amount, and the processing amount in execution for each process, and the distribution unit specifies to which process the received request is a request. Then, a server to which the request is assigned may be determined based on the priority for the identified process, the maximum processing amount, and the in-execution processing amount.

  In a preferred embodiment, each of the plurality of servers notifies the load balancer of processes that can be executed by the server, and the load balancer assigns a priority for each process to the server that has received the notification. The information processing apparatus may further include means for assigning and registering in the storage means.

  In a preferred embodiment, the priority order may be set higher for a process executed in a later server in the connection order to the load balancer.

  In a preferred embodiment, each of the plurality of servers may notify the load balancer of the maximum processing amount for each process that can be executed by the server.

  In a preferred embodiment, the maximum processing amount for each process is a maximum number of requests that can be requested to each server for each process, and the in-execution processing amount is a process in which processing is executed in each server. It may be a different number of requests.

  According to an embodiment of the present invention, a load distribution apparatus that distributes requests to any of a plurality of servers, a storage unit that stores a priority order, a maximum processing amount, and an executing processing amount for each of the plurality of servers; When a request is received, the storage unit is referred to, and a unit that allocates the request to a server with the highest priority among servers in which the processing amount during execution does not reach the maximum processing amount is provided.

  Hereinafter, a load distribution processing system according to an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram illustrating a configuration of a load distribution processing system according to the present embodiment.

  This system receives requests from a plurality of clients 5 connected via a network 9 and executes a load distribution device 1 assigned to any one of the servers 3 and a process corresponding to the request assigned from the load distribution device 1 A plurality of servers 3.

  Each of the load balancer 1, each server 3, and each client 5 is configured by, for example, a general-purpose computer system. Individual components or functions in the load balancer 1, each server 3, and each client 5 described below. Is realized, for example, by executing a computer program.

  Also, the specifications of each server 3 such as the model and processing capacity of each server 3 do not have to be the same, and may be different from each other.

  Each client 5 generates and outputs a request for requesting execution of a predetermined process. This request includes information (for example, process name) indicating to which process operating on the server 3 the request.

  Each server 3 notifies the load balancer 1 of its own identification information such as a server name and process identification information such as an executable process name. For example, when a certain server 3 is newly connected to the load balancer 1 or when a new process is added to the already connected server 3, those servers 3 have the above server name and process. You may make it notify of a name.

  The load balancer 1 receives a request from the client 5 and assigns the request to any of the plurality of servers 3, a storage unit 13 that stores the reference table 18 and the status table 19, and a storage unit 13 includes a registration processing unit 15 that registers the reference table 18.

  FIG. 2 shows an example of the reference table 18.

  In the present embodiment, the reference table 18 is configured by a process-specific table, as shown in FIG. The example in the figure is a reference table 18 for process a. The reference table 18 includes a server name 181 that can be executed by the target process, a priority 182 when assigning the process of the target process to each server, and a maximum process that can be requested to each server with respect to the target process. The maximum request number 183 indicating the amount is included.

  In the status table 19, the execution state of the process in each server 3 is registered.

  FIG. 3 shows an example of the status table 19.

  As shown in the figure, the status table 19 is configured by a process-specific table, similar to the reference table 18. The example of FIG. 6 is a status table 19 for process a. The status table 19 includes a server name 191 and the number of currently executing requests 192 indicating the amount of processing executed by each server for the target process.

  When the registration processing unit 15 receives a notification of a process name executable on each server from each server 3, the registration processing unit 15 registers this information in the reference table 18. For example, upon receiving notification from the server A that “process a” and “process b” are executable, the server A is registered in the reference table 18 of “process a” and “process b”, respectively.

  The priority order 182 may be fixedly set in advance. However, if the priority order 182 is dynamically set according to the specification or state of the server 3, a failure occurs in any of the servers 3. Does not affect the entire system and does not require maintenance. As an example, the priority order 182 is automatically set by the registration processing unit 15 of the load distribution apparatus 1 according to the connection order of the servers 3 to the load distribution apparatus 1 (starting order of the servers 3). . If the priority order 182 at this time is set so that the connection order to the load balancer 1 becomes higher as the process executed by the server 3 later, the server newly added according to the shortage of the processing capacity of the entire system The ability of 3 can be fully utilized.

  Further, the maximum number of requests 183 for each process may be fixedly set in advance, but it is desirable to notify each server 3 together with the process name notification described above. This is because the maximum processing load for each server 3 is determined according to the maximum number of requests, and it is preferable that each server 3 designates.

  When the request processing unit 11 receives a request from any of the clients 5, the request processing unit 11 refers to the reference table 18 and the status table 19, assigns the request to any one of the servers 3, and executes a process corresponding to the request.

  For example, when receiving a request, the request processing unit 11 detects a process name included in the request and specifies a process for executing the request. Then, by referring to the reference table 18 and the status table 19 related to the process, the server having the highest priority 182 among the servers in which the number of executing requests 192 has not reached the maximum number of requests 183 is identified, and a request is assigned to the server assign.

  For example, when receiving a request for the process a, the request processing unit 11 refers to the reference table 18 and the status table 19 of the process a illustrated in FIG. It is determined whether the number 192 has reached the maximum number of requests 183. A request is assigned to the server 3 having the highest priority among the servers 3 in which the number of executing requests 192 does not reach the maximum number of requests 183.

  In the example of FIG. 2, the server A having the highest priority for the process a has the number of requests 192 already being executed reaching the maximum number of requests 183. Therefore, referring to the server B with the next highest priority, since the number of requests 192 being executed has a margin up to the maximum number of requests 183, the server B is specified as the allocation target server. Then, the request processing unit 11 causes the server B to execute the accepted request.

  The request processing unit 11 further manages the status table 19. That is, when the request processing unit 11 newly allocates a request, 1 is added to the number of executing requests 192 of the allocated server 3. When the processing related to the allocated request is completed, 1 is subtracted from the number of requests 192 being executed by the server 3.

  As a result, the request processing unit 11 continuously assigns processes to the server 3 until the number of executing requests 192 of the server 3 with the highest priority reaches the maximum number of requests 183. When the number of requests 192 in execution of the server 3 with the highest priority reaches the maximum number of requests 183, the number of requests 192 in progress similarly reaches the maximum number of requests 183 for the server 3 with the next highest priority. Until then, the server 3 is continuously requested to process. This is repeated thereafter.

  Note that when the processing is completed in any one of the servers 3, the number of executing requests 192 is also updated accordingly. Therefore, for example, even when the server with the highest priority reaches the maximum number of requests 183 once, the number of requests being executed 192 may be less than the maximum number of requests 183, resulting in a so-called free state. When a new request is received at this time, assignment is performed according to the above-described processing procedure, so that the request is assigned to the highest server.

  In this way, when load balancing is realized by requesting processing until the upper limit of the processing amount that can be requested is reached in order from the server with the highest priority, the processing capacity by process is the entire load balancing processing system. It is possible to determine the load status of the entire system very easily. For example, by referring to the number of requests 192 being executed by the server having the lowest priority, it can be easily determined whether the processing capacity of the process is insufficient in the entire load distribution processing system. In other words, if the number of requests being executed on the server with the lowest priority is the maximum request number that is the upper limit value or a number close to it, the load on the entire system is heavy and the processing capacity is insufficient or insufficient. It can be seen that it is.

  Further, by referring to the number of requests 192 being executed for all the servers, it is possible to easily determine how much of the processing capacity of the entire system is used.

  The above-described embodiments of the present invention are examples for explaining the present invention, and are not intended to limit the scope of the present invention only to those embodiments. Those skilled in the art can implement the present invention in various other modes without departing from the gist of the present invention.

  For example, in the above-described embodiment, load distribution is performed based on the priority order for each process, and it is determined whether the processing capability is sufficient for each process. However, load distribution similar to the above may be performed without dividing each process.

  In the above-described embodiment, the number of requests is used as a parameter for measuring the maximum processing amount and the processing amount during execution. However, other parameters such as CPU time may be used.

1 shows a configuration of a load distribution processing system according to an embodiment of the present invention. An example of the reference table 18 is shown. An example of the status table 19 is shown.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Load distribution apparatus 3 Server 5 Client 11 Request processing part 13 Storage part 15 Registration processing part 18 Reference table 19 Status table

Claims (11)

A load balancing processing system comprising a load balancing device and a plurality of servers,
Each of the servers can execute a plurality of processes.
The load balancer is:
In order to assign a request from a client to each server, the priority of the server fixedly set for each process, the maximum number of requests that can be requested to each server with respect to the target process, and the target Storage means for storing the number of requests currently being executed by each of the servers with respect to the process ,
Each time a request is received from a client , a process that is a target of the received request is specified, and the number of requests being executed does not reach the maximum number of requests with reference to the storage unit for the target process. A target server that is a server having the highest priority among the servers is set, and requests are continuously allocated to the target server until the number of requests being executed reaches the maximum number of requests. When the number of requests being executed reaches the maximum number of requests , the target server is set to the next highest priority server, so that the target process is sequentially performed from the higher priority server. swing for assigning the request to reach the maximum number of requests Only load distribution processing system comprising a means. A load balancing processing system comprising a load balancing device and a plurality of servers,
Each of the plurality of servers can execute a plurality of processes.
The load balancer is:
A storage means for storing a priority for each server indicating a priority for assigning a process to each server, a maximum processing amount in each server, and a processing amount in execution in each server for each process;
When receiving a request, the storage means is referred to, and a distribution means for allocating the request to a server with the highest priority among servers in which the processing amount during execution does not reach the maximum processing amount,
The distribution unit specifies a process to which the received request is a request, and assigns the request based on the priority order, the maximum processing amount, and the processing amount being executed for the specified process. Decide
The load distribution processing system, wherein the priority is set higher for a process executed by a server having a later connection order to the load distribution apparatus. The maximum processing amount for each process is the maximum number of requests that can be requested to each server for each process,
The load distribution processing system according to claim 2, wherein the processing amount during execution is the number of requests for each process executing processing in each server. Each of the plurality of servers notifies the load balancer of processes that can be executed by the own server,
4. The load distribution processing system according to claim 1, wherein the load distribution apparatus further includes means for assigning a priority for each process to a server that has received the notification and registering the priority in the storage unit. 5. 5. The load distribution processing system according to claim 1, wherein each of the plurality of servers notifies the load distribution apparatus of the maximum processing amount or the maximum number of requests for each process that can be executed by the server itself. . A load balancer that distributes requests from clients to any of a plurality of servers capable of executing a plurality of processes ,
And fixedly set priority order of the respective servers for each of the processes, the maximum number of requests that can be requested to the each server with respect to the process of interest, each server with respect to the process to be the target is currently executing Storage means for storing the number of requests being executed ;
Each time a request is received from a client, the target process of the received request is identified, and the storage unit is referred to , and the number of requests being executed for the target process does not reach the maximum number of requests. A target server that is the highest priority server is set, and requests are continuously allocated to the target server until the number of requests being executed reaches the maximum number of requests, and the execution of the target server is performed. If the number of middle requests has reached the number of the maximum request, said that the next priority target server is set to high server, with respect to the process to be the target, sequentially from higher the priority the server, the sorting for assigning the request to reach maximum number of requests Load balancer comprising: a stage, a. A load balancing method for a load balancer to distribute requests from clients to any of a plurality of servers capable of executing a plurality of processes ,
The load balancer is:
A step of storing the priority fixedly configured server, the maximum number of requests and possible requests to the each server with respect to processes of interest for each of the process,
Storing the number of requests currently being executed by each of the servers with respect to the target process;
Each time a request from a client is received, the process that is the target of the received request is identified, and the number of requests being executed for the target process reaches the maximum number of requests with reference to the storage unit. A target server that is the highest priority server among the servers that do not exist is set, and requests are continuously allocated to the target server until the number of requests being executed reaches the maximum number of requests . When the number of requests being executed reaches the maximum number of requests , the target server is set as the server with the next highest priority, so that the target process is sequentially performed from the server with the highest priority. , scan of allocating the request until it reaches the maximum number of requests Load balancing method comprising Tsu and up, the. A computer program for load balancing that distributes requests from clients to any of a plurality of servers capable of executing a plurality of processes ,
When executed on a computer,
A step of storing a fixedly set priority order of each server, the maximum number of requests and possible requests to the each server with respect to the process of interest, to each of the process,
Storing the amount of processing being executed in each of the plurality of servers;
Each time a request from a client is received, the process that is the target of the received request is identified, and the number of requests being executed for the target process reaches the maximum number of requests with reference to the storage unit. A target server that is the highest priority server among the servers that do not exist is set, and requests are continuously allocated to the target server until the number of requests being executed reaches the maximum number of requests . When the number of requests being executed reaches the maximum number of requests , the target server is set as the server with the next highest priority, so that the target process is sequentially performed from the server with the highest priority. , scan of allocating the request until it reaches the maximum number of requests A computer program that Tsu and up, are carried out. A load balancer that distributes requests to any of a plurality of servers,
Each of the plurality of servers can execute a plurality of processes.
For each of the plurality of servers, storage means for storing priority, maximum processing amount, and executing processing amount for each process;
When accepting a request, the storage means is referred to, and means for allocating the request to a server with the highest priority among servers in which the processing amount during execution does not reach the maximum processing amount,
The means for allocating the request specifies to which process the received request is a request, and assigns the request based on the priority order, the maximum processing amount, and the processing amount being executed for the specified process. Decide which server to assign,
The priority is set higher for a process executed by a server having a later connection order to the load balancer. A load balancing method for a load balancer to distribute requests to any of a plurality of servers,
Each of the plurality of servers can execute a plurality of processes.
The load balancer is:
For each of the plurality of servers, a priority order and a maximum processing amount are stored for each process;
Storing the amount of processing being executed in each of the plurality of servers by process;
A request is received, and the request is allocated to a server with the highest priority among servers in which the processing amount being executed does not reach the maximum processing amount,
The step of assigning the request specifies to which process the received request is a request, and assigns the request based on the priority order, the maximum processing amount, and the processing amount in execution for the specified process. Decide which server to assign,
The load distribution method, wherein the priority is set higher for a process executed on a server with a later connection order to the load distribution apparatus. A computer program for load balancing that distributes requests to any of a plurality of servers,
Each of the plurality of servers can execute a plurality of processes.
When executed on a computer,
For each of the plurality of servers, storing a priority and a maximum processing amount for each process;
Storing the amount of processing being executed in each of the plurality of servers by process;
When a request is accepted, a step of allocating the request to a server having the highest priority among servers in which the processing amount being executed has not reached the maximum processing amount is performed,
The step of assigning the request specifies to which process the received request is a request, and assigns the request based on the priority order, the maximum processing amount, and the processing amount in execution for the specified process. Decide which server to assign,
The computer program in which the priority is set higher as a process is executed in a later server in the connection order to the load balancer.

Images