WO2004084085A1

WO2004084085A1 - Load distributing system by intersite cooperation

Info

Publication number: WO2004084085A1
Application number: PCT/JP2003/003273
Authority: WO
Inventors: Tsutomu Kawai; Satoshi Tutiya; Yasuhiro Kokusho
Original assignee: Fujitsu Limited
Priority date: 2003-03-18
Filing date: 2003-03-18
Publication date: 2004-09-30
Also published as: JPWO2004084085A1

Abstract

A system comprises a front-stage center (12-1) for directly receiving a request from a client (10) through a network (11) and a back-stage center (12-2) for receiving the request from the client (10) through the front-stage center (12-1). The centers have auxiliary servers (17-1, 17-2), respectively. The front-stage center (12-1) provides a service by using a normal server. A system controller (16-1), on detecting that the load on the server increases, provides a server for providing the service the load of which increases from the auxiliary server (17-1) commonly provided for a service 1 and a service 2. If the load cannot be supported even by the provision of the server, the system controller (16-1) issues an instruction to a system controller (16-2) of the back-stage center (12-2) to support the provision of the service. When the back-stage controller (12-2) cannot support the load by using a normal server, it supports the load by using the auxiliary server (17-2).

Description

Description Load distribution system by linking sites

The present invention relates to a load distribution system using inter-site cooperation. Background art

Due to the explosive spread of the Internet, resources such as servers and networks required by service providers have become enormous. However, it is known that the amount of requests from users fluctuates greatly depending on time and conditions.If resources are secured at the time of concentration, it is usually necessary to maintain useless resources at normal times. In some cases, resources that cannot be handled can degrade service quality and give users discomfort. Furthermore, as the number of users increases, it becomes difficult to estimate the upper limit of the required resources, and a system that allocates resources as needed becomes necessary. At the same time, excess resources can lead to increased management costs, and there is also a need for more efficient use of unnecessary resources.

Figure 1 shows an example of a conventional load distribution system.

In the configuration of FIG. 1, the client 10 accesses the data center 12 via the network 11 and receives a service. A plurality of servers 14 are connected to the load balancer 13.

If a single server cannot handle the process, install multiple servers as shown in Fig. 1 and place a load balancer 13 in front of it to distribute the load to multiple servers and improve service quality. However, additional determination of server 14 and server 14 load balancing In many cases, the work of adding the device 13 and changing the setting is performed manually, and it is necessary to secure a server at all times corresponding to the maximum load, resulting in a large cost.

Patent Document 1 defines a method of adding a server and distributing requests from users. However, it is necessary to incorporate a mechanism for server selection on the user side, which is suitable for application to an unspecified number of services. Not. In addition, there is a problem that it is necessary to exchange management information other than the request.

Further, the method of Patent Document 2 can be applied only to a case where static information is distributed, and cannot be applied to a case where different information is returned every time in response to a request from a user such as service provision.

Furthermore, Patent Document 3 also assumes the case of static information, and does not consider the case where the load on the file server or the like becomes excessive.

Patent Document 1

JP-A-9-1106381

Patent Document 2

Japanese Patent Application Laid-Open No. Hei 91-179820

Patent Document 3

JP 2002-259354 A DISCLOSURE OF THE INVENTION

An object of the present invention is to provide a load distribution system capable of distributing a load for providing a service and flexibly responding to a change in a request from a user.

The method of the present invention is a method of distributing the load of an apparatus having a plurality of servers for providing a service to a client via a network. Providing a plurality of spare servers that are not set up for normal services; In anticipation of an increase in server load, an abbreviated setting for a service to be provided to the spare server is set, and the server for providing the service is provided. The control step for sharing the load with the server that normally provides the service is performed. It is characterized by having. According to the present invention, in a device such as a data center, a plurality of spare servers are provided in addition to a server for providing a normal service, and when a load on a server for providing a normal service increases, the spare server is provided. Then, install an abridge so that the service can be provided, and share the load of the server for providing the service.

In another aspect, according to the present invention, devices equipped with spare servers are connected via a network, and control is performed so that spare servers are provided to each other. Even if it does not have enough processing power to support the service, multiple devices can cope with the load via the network and cope with the load, thereby avoiding interruption of service provision due to a large load. In addition, this allows the number of spare servers to be provided for one device to be reduced, eliminating the need for redundant hardware in each device. BRIEF DESCRIPTION OF THE FIGURES

Figure 1 shows an example of a conventional load distribution system.

FIG. 2 is a diagram showing a basic configuration of the embodiment of the present invention.

FIG. 3 is a diagram showing a network arrangement configuration in a center in the basic configuration of FIG.

FIG. 4 is a diagram showing a first embodiment of the present invention.

FIG. 5 is a diagram illustrating the operation of the first exemplary embodiment of the present invention.

FIG. 6 is a diagram showing data for calculating the load and capacity of the server. Figure 7 shows data for selecting a server according to the size of the load. is there.

FIG. 8 is a diagram showing the relationship between the capacity of the server to be added and the predicted value of the load. FIG. 9 is a diagram showing a configuration in which a spare server is shared by a plurality of services. FIG. 10 is a diagram showing a configuration in a case where a spare server is provided between different centers.

FIG. 11 is a diagram illustrating the operation of the embodiment of the present invention.

FIG. 12 is a diagram for explaining how to secure a network band when cooperating with another center.

FIG. 13 is a diagram illustrating an application example of the embodiment of the present invention in a web server. FIG. 14 is a diagram showing an application example of the embodiment of the present invention in a web service. FIG. 15 is an application example of the embodiment of the present invention in a case where equal centers mutually exchange resources. is there.

FIG. 16 is a diagram showing an example in which the embodiment of the present invention is applied to a pre-center having no spare server.

FIGS. 17 to 24 are flowcharts illustrating the operation of the embodiment of the present invention when there is no cooperation between databases provided in the center.

FIG. 25 to FIG. 30 are flowcharts showing the processing flow of the embodiment of the present invention when the database is linked. BEST MODE FOR CARRYING OUT THE INVENTION

In the present invention, service quality is assured by predicting changes in the amount of requests from users, and dynamically adding and deleting servers in the data center or other linked data centers in accordance with the change. The aim is to reduce costs by sharing the surplus server with multiple services. FIG. 2 is a diagram showing a basic configuration of the embodiment of the present invention.

The client 10 accesses the web server 15-1 via the network 11 via the load distribution device 13_1 of the preceding center 12-1. As a result of the data processing in the web server 15-1, the client 10 accesses the database server 14-1 or the file server 14-12 to receive the service. The rear-stage server 1 2-2 has almost the same configuration as the front-stage center 12-1, receives a request from the client 10 via the load balancer 13-1, and loads the load balancer 13-2 The client 10 is guided to the web server 15-2 while distributing the load with. Then, the client 10 accesses the database server 14-3 or 14_4 via the web server 15-2 to receive the service.

Here, the first-stage center 12-1 indicates a center that directly receives a user's request, and the second-stage center 12-2 indicates a center that processes a user request through the first-stage center 12-1. The assignment of servers between data centers is a many-to-many relationship, such as when a certain data center uses servers from multiple data centers or when a certain data center responds to server requests from multiple data centers simultaneously. is there. Server load status ゃ Client load status is determined by the system controller 1 6-1,

16-2 performs the tallying / judgment 'application, and sets the results in the servers 141-14-14-14 and the load balancers 13-1 and 13-2. If server resources are insufficient, set servers 17_1 and 17-2 in the spare server as servers with necessary functions, add them to the service, and improve their performance.

The physical network configuration consists of connecting all the servers directly under a single switch group 20, and having a logically independent network (VLAN0, VLAN11, VLAN12, VLAN21). With such an arrangement, it becomes possible to automate the process of adding servers to the required locations.

When adding or deleting servers, the server capacity is derived from server specifications such as CPU performance / network configuration, and necessary servers are calculated even in an environment where various types of hardware are mixed. Assign servers appropriately. At the same time, it calculates the traffic to that server and secures or arbitrates the network bandwidth.

In addition, by predicting the future load from load measurement and load fluctuation prediction, servers will be added before an overload occurs, and service quality will be guaranteed.

FIG. 4 is a diagram showing a first embodiment of the present invention.

In the figure, the same components as those in FIG. 2 are denoted by the same reference numerals, and detailed description is omitted.

If the user's request exceeds the capacity of the assigned server, the response time will increase or no response will occur, which will make the user uncomfortable. If the load further increases in this state, a server failure may be caused. To prevent this, the system control unit 16 measures the load status of the server, and if it is judged that the current number of servers will cause a problem, a server is added from the spare server 17 and applications, services, Set and introduce data to be used. Then, the settings of the dependent devices and servers are updated and incorporated into the service.

In the figure, the same components as those in FIG. 4 are denoted by the same reference numerals, and description thereof will be omitted.

When the amount of requests from users decreases, surplus servers are generated. Even if this surplus server is deleted, the service quality does not decrease. It is desirable to open it as a spare server and use it for other services from the viewpoint of operating cost / improvement of utilization rate. For this reason, by removing the related settings from the dependent devices, the coordination with the service is canceled, and then the processing such as the release of the settings is performed, and then the spare server 17 is returned.

FIG. 6 is a diagram showing data for calculating the load and capacity of the server. In order to add and remove service capabilities as needed, information is required on how many service capabilities a given server provides. In a data center, etc., the service capacity per unit changes depending on the combination of servers and devices used, and applications and services. It is practically impossible to use uniform servers when multiple data centers cooperate.Therefore, it is necessary to calculate service capabilities from equipment specifications such as CPU and memory. is there. Therefore, from the performance value in a typical configuration,

A method of estimating a performance value in consideration of a difference in CPU capability and the like is used.

Figure 7 is a diagram showing data for selecting a server according to the size of the load.

Here, not only the service capability but also information on what kind of use it is preferable to use as a characteristic of the server unit is retained. As described above, the available performance values for each server are not uniform, so it is necessary to create a configuration that can provide the required capabilities by combining these. For this reason, from the performance values and characteristics obtained in Fig. 6 and the required performance values, the server with the higher recommendation is preferentially selected and used until the required amount is satisfied.

FIG. 8 is a diagram showing the relationship between the capacity of the server to be added and the predicted value of the load. Simply adding resources when the measured request volume exceeds the service capacity cannot guarantee service quality when the load is rapidly increasing. For this reason, the trend of the load is grasped, and when the request volume is expected to increase, Prevents service quality from deteriorating by adding in advance a service capability that matches the expected request volume. As a method of prediction, linear extrapolation and the like can be considered.

FIG. 9 is a diagram showing a configuration in which a spare server is shared by a plurality of services. When looking at the load status of multiple services in a data center, it is extremely rare that all services have a high load at the same time.If a spare resource is secured for each service, there will always be unused resources. Conceivable. By sharing spare resources among multiple services, it is possible to add the required service capacity with less spare resources as a whole. Also, the maintenance cost can be dispersed by sharing. The service center 1 and service 2 are equipped with the service 1 and the service 2, respectively, and the load balancers 13-1 and 13-2 are provided respectively. Service 1 has a Web server 15-1, a database server 14-1, and a file server 14-2. The service 2 is provided with a server 25. The spare server 17 is provided in common for service 1 and service 2.The system controller 16 checks the load status and installs additional servers from spare server 17 to service 1 or service 2. I do.

FIG. 10 is a diagram showing a configuration in a case where a spare server is provided between different centers.

In the figure, the same components as those in FIG. 2 are denoted by the same reference numerals, and description thereof will be omitted.

Depending on the size of the data center 12-1, there may be cases where a sufficient spare server 17-1 cannot be secured physically or cost-effectively even if the spare server is shared between different services. Even if it is intended to be sufficient, a sudden load may not be able to cover the spare server in the data center. In such a case, another data The center 12-2 is a post-center and its spare server 17-2 is used via the network.

In the figure, the same components as those in FIG. 9 are denoted by the same reference numerals, and description thereof will be omitted.

Some services require servers that cooperate with each other, such as databases, in addition to servers that directly exchange information with users. In the case of such a service, the performance cannot be improved unless the processing capacity and load status are checked for each function and a server is added to an appropriate function. For this reason, the system controller 16 checks the load for each layer, and when adding or deleting, changes the setting of the linked server to increase or decrease the capacity.

In the figure, the same components as those in FIG. 10 are denoted by the same reference numerals. When multiple services operate simultaneously or when cooperative processing is required, sufficient processing capacity cannot be obtained unless servers are added and traffic between services and functions is arbitrated. Calculate the required bandwidth in each part, and secure each bandwidth to the network in consideration of the ratio so that sufficient performance can be obtained as a whole.

With the above configuration, the load from the user and the status of the server capacity can be monitored, and the necessary and sufficient resources can be allocated from the data center or the linked data center before the load exceeds the server capacity. Therefore, it is possible to guarantee service quality for requests from users. Since the required spare servers can be shared widely over a wide area, the total number of required servers can be reduced as a whole. In addition, a service in which servers with multiple functions cooperate Server, it is possible to add a server to the function that is the bottleneck, so it is possible to achieve a sufficiently large scale. In addition, since the entire process can be automated, it can quickly follow changes in the amount requested by the user.

FIG. 13 is a diagram illustrating an application example of the embodiment of the present invention in a web server. In the figure, the same components as those in FIG. 12 are denoted by the same reference numerals, and description thereof will be omitted.

When the load is light, only the front center 12-1 is operated. When the load increases, the spare server 17-1 in the preceding center 12-1 is added as the web server 15-1. When the load further increases, a web server group 15-2 is created in the rear center 12-2 so that the rear center 12-2 can handle the load. FIG. 14 is a diagram showing an application example of the embodiment of the present invention in a web service.

In the figure, the same components as those in FIG. 12 are denoted by the same reference numerals, and description thereof will be omitted.

In this example, the web service is composed of a combination of a web server 15-1, a database server 14-1, and a file server 14_2. When the load is light, only the front center 12-1 is operated. As the load increases, spare servers 17-1 are sequentially added to the bottleneck, and if the server cannot be cut off by the preceding center 12-1, the spare server 17-1 is connected to the latter. Cooperate. In this example, the database server 14-1 synchronizes data even during cooperation between the preceding center 12_1 and the following center 12-2. This is realized by creating a V lan across the centers and securing the bandwidth.

FIG. 15 shows an application example of the embodiment of the present invention when equal centers mutually exchange resources.

The processing capacity of service 1 in center 1 is less than spare server 3 0—1 in center 1. In the event of failure, request cooperation from Center 2 and use the servers in Center 2 (shaded area and spare server 30-3). Further, when the server capacity in the center 2 is also dead (when the capacity including the spare server 30-2 is dead), another center 3 is requested to cooperate, and the server in the center 3 is shaded (shaded). Partial and spare servers 30-3) are used.

When the system control unit 16-1 determines that there is not enough server for service provision in the first center 12-1, the second center 12-2 is requested to cooperate, and the second center 12-2 is requested. Use the server in 2. Here, a load balancer and a Web server are provided for service 1 and service 2. The service 1 and service 2 servers provide service 1 and service 2, respectively. Further, in the latter-stage center 12-2, when the capacity of the server becomes insufficient, the spare server 17 is added as needed for each service. The system control unit 16 _ 2 determines the addition and cooperates with the preceding center 12-1.

FIGS. 17 to 24 are flowcharts for explaining the operation of the embodiment of the present invention in the case where the databases provided in the center are not linked.

FIG. 17 is a flowchart showing the overall flow of the system control device. First, in step S10, load measurement is performed. In step S11, it is determined whether the predicted processing capacity exceeds the allocated processing capacity. If the determination in step S11 is YES, in step S12, the processing capacity is added, and the process proceeds to step S15. In step S15, the wait is 10 seconds. Force This value should be set by the designer as appropriate.

If the determination in step S11 is NO, in step S13, it is determined whether the current processing capacity is less than half of the allocated processing capacity. If the determination in step S13 is YES, in step S14, the processing capacity is reduced, and the process proceeds to step S15. If the determination in step S13 is NO, the process proceeds to step S15.

After step S15, the process returns to step S10 again.

FIG. 18 is a diagram showing the details of the load measurement in step S10 of FIG. In step S20, the average number of processes for 10 seconds is collected from the server in use. This 10 seconds should match the value of step S15 in FIG. In step S21, the total average number of processes is calculated and added to the measurement history. In step S22, it is determined whether there are four or more measurement histories. If the determination in step S22 is NO, in step S23, the latest history is set as a predicted value 30 seconds later, and the process proceeds to step S25. If the determination in step S22 is YES, in step S24, a predicted value 30 seconds later is calculated from the last four histories by least squares approximation, and the process proceeds to step S25. This means finding a regression curve from the latest four histories and using the regression curve to obtain a predicted value 30 seconds later. In step S25, a predicted value after 30 seconds is set. In step S26, the latest history is set to the current value, and the process returns to the flow in FIG.

FIG. 19 is a diagram showing the details of the processing capacity adding process in step S12 of FIG.

In step S30, the current processing value is subtracted from the predicted value to determine the additional processing capacity. In step S31, it is determined whether there is a spare server in the center. If the determination in step S31 is YES, in step S32, an additional server in the center is selected. In step S33, it is determined whether or not the additional processing capacity has been satisfied. If the determination in step S33 is N〇, the flow proceeds to step S34, and if the determination is YES, the flow proceeds to step S38. If the determination in step S31 is NO, the process proceeds to step S34. At 34, it is determined whether or not there is a partner center having a preliminary processing capacity. If the determination in step S34 is YES, in step S36, the coordination center allocates processing capacity. In step S37, it is determined whether or not the additional processing capacity has been satisfied. If the determination in step S37 is NO, the process proceeds to step S34. If the determination in step S37 is YES, the process proceeds to step S38. If the determination in step S34 is NO, in step S35, the administrator is warned that the additional processing capacity cannot be satisfied, and the process proceeds to step S38. In step S38, a VLAN is set so as to include the selected server. In step S39, an application is set for the selected server, and the process proceeds to step S40.

In step S40, it is determined whether or not there is cooperation between the centers. If the determination is NO, the process proceeds to step S43. If the determination in step S40 is YES, in step S41, the coordination center load distribution ratio is determined and assigned, and the equipment is set. In step S42, the own center and the coordination center are connected. Set the communication band between them, and proceed to step S43. In step S43, the load distribution ratio of the own center is determined, the allocating device is determined, and the process returns to the flow of FIG.

FIG. 20 is a flow showing in detail the process of selecting an additional server in step S32 of FIG.

In step S50, it is determined whether there is a server for a necessary use. If the determination in step S50 is NO, the process proceeds to step S54. If the determination in step S50 is YES, in step S51, it is determined whether or not there is a server capable of satisfying the additional processing capacity with a single server for the required application. If the determination in step S51 is NO, in step S52, the server with the highest performance for the required application is selected, and the process returns to step S50. Steps If the determination in S51 is YES, the server with the lowest performance is selected from among the servers for the required applications, which can provide the additional processing capacity, and the process proceeds to step S58. .

In step S54, it is determined whether there is an available server. If the determination in step S54 is NO, the process proceeds to step S58. If the determination in step S54 is YES, in step S55, it is determined whether one server can satisfy the additional processing capacity. If the judgment in step S55 is NO, in step S56, the server with the highest performance is selected, and the process returns to step S54. If the determination in step S55 is YES, in step S57, the server with the lowest performance is selected from the servers that can satisfy the additional processing capacity with one, and the process proceeds to step S58. move on. In step S58, a list of assigned servers is configured, and the process returns to the process in FIG.

FIG. 21 is a flowchart showing the flow of the coordination center processing capacity assignment processing in step S36 in FIG.

In step S60, it is determined whether or not the processing capacity upper limit based on the bandwidth is smaller than the desired allocation value. If the determination in step S60 is NO, the process proceeds to step S62. If the determination in step S60 is YES, in step S61, the upper limit of the quota is set as the upper limit of the bandwidth, and the process proceeds to step S62.

In step S62, a request is made to the partner center to select a server. In step S63, an additional server is selected in the partner center, and in step S64, a list of assigned servers is configured. Then, the processing returns to the processing in FIG.

FIG. 22 is a detailed flow of the application setting in step S39 in FIG.

In step S70, it is determined whether or not there is cooperation between the centers. If the determination in step S70 is NO, the process proceeds to step S74. Step S 70 If the determination is YES, in step S71, it is determined whether or not the application live has been transferred. If the determination in step S71 is YES, the process proceeds to step S73. If the determination in step S71 is NO, in step S72, the application archive is transferred to the partner center, and the process proceeds to step S73. In step S73, the application is installed on the additional server, and the process proceeds to step S74. In step S74, the application is installed on the additional server in the own center, and the process returns to the process in FIG. FIG. 23 is a flowchart showing the processing for reducing the processing capacity in step S14 of FIG.

In step S80, the current measured value is subtracted from the assigned value to determine the reduction processing capacity. In step S81, it is determined whether there is a cooperation center. If the determination in step S81 is YES, in step S82, a reduction server is determined in the cooperation center, and in step S83, it is determined whether all servers in the cooperation center have been reduced. If the determination in step S83 is YES, the process returns to step S81. If the determination in step S83 is NO, the process proceeds to step S85. If the determination in step S81 is NO, in step S84, the own server determines the reduction server, and the process proceeds to step S85.

In step S85, the load distribution ratio of the own center is determined, and the allocation device is set. In step S86, the load distribution ratio of the cooperation center is determined, and the assignment device is set. Then, in step S87, the process waits for completion of the user request process. In step S88, the application is deleted from the reduction server, and in step S89, a VLAN is set so as to include only the remaining servers (coordination network communication path is set). In step S90, It is determined whether or not the cooperation is released. If the determination in step S90 is YES, in step S91, the bandwidths of the own center and the cooperation center are released and Return to the processing of step 7. When the determination in step S90 is NO, the process returns to the process in FIG. FIG. 24 is a flowchart showing the selection processing of the reduction server in step S82 or step S84 in FIG.

In step S100, it is determined whether there is a server available for another use. If the determination in step S100 is NO, the process proceeds to step S103. If the determination in step S100 is YES, in step S101, it is determined whether there is a server whose performance is lower than the remaining reduction capacity. If the determination in step S101 is NO, the process proceeds to step S103. If the determination in step S101 is YES, in step S102, among the servers whose performance is lower than the remaining reduction performance, the server with the highest performance is reduced, and the process proceeds to step S100.

In step S103, it is determined whether there is a server currently in use. If the determination in step S103 is NO, the process proceeds to step S106. If the determination in step S103 is YES, in step S104, it is determined whether there is a server whose performance is lower than the remaining reduction performance. If the determination in step S104 is NO, the process proceeds to step S106. If the determination in step S104 is YES, in step S105, the server with the highest performance among the servers having lower performance than the remaining reduction performance is reduced, and the process returns to step S103.

In step S106, a list of the deleted servers is generated, and the process returns to the process in FIG.

FIG. 25 to FIG. 30 are flowcharts showing the processing flow of the embodiment of the present invention when there is cooperation of databases.

FIG. 25 is a flowchart showing the flow of the overall processing of the own center that performs the cooperation request. In step S110, the load of the Web server is measured. In step S111, it is determined whether the predicted processing capacity is larger than the allocated processing capacity. If the determination in step S111 is YE S, in step S112, Web processing capacity is added, and the process proceeds to step S115. If the determination in step S111 is NO, in step S113, it is determined whether the current processing capacity is smaller than one half of the allocated processing capacity. If the judgment in step S113 is N〇, the process proceeds to step S115. If the determination in step S113 is YES, in step S114, the capacity of the Web processing is reduced, and the process proceeds to step S115. In step S115, the load on the database in the center is measured. In step S116, it is determined whether the predicted processing capacity is larger than the allocated processing capacity. If the determination in step S116 is YES, in step S117, the database processing capacity is added, and the flow advances to step S120. If the determination in step S116 is NO, in step S118, it is determined whether the current processing capacity is smaller than one half of the allocated processing capacity. If the determination in step S118 is NO, the process proceeds to step S120. If the determination in step S118 is YES, in step S119, the processing capacity of the database is reduced, and the flow advances to step S120. In step S120, the process waits for 10 seconds. This waiting time should be set as appropriate by the designer. After step S120, again go to step S110

Fig. 26 is a flow chart showing the overall processing flow of the partner center.

In step S130, the database load in the center is measured. In step S131, it is determined whether the predicted processing capacity is larger than the allocated processing capacity. If the determination in step S 13 1 is YES, in step S 13 2, the database processing capacity is added, and the flow advances to step S 1 35. If the determination in step S133 is NO, in step S133, it is determined whether the current processing capacity is smaller than one half of the allocated processing capacity. If the determination in step S133 is NO, the process proceeds to step S135. Step S 1 3 3 If the determination is YES, the database processing capacity is reduced in step S134, and the flow advances to step S135. In step S135, the process waits for 10 seconds, and returns to step S130. This 10 seconds should not be limited to this, but should be set appropriately by the designer.

Figure 27 is a flowchart showing the detailed processing of web load measurement or database load measurement performed at each center.

In step S140, the average number of processes for 10 seconds from the server in use is collected. This 10 seconds should be the same value as the waiting time of step S120 of FIG. 25 and step S135 of FIG. In step S141, the total average number of processes is calculated and added to the measurement history. In step S142, it is determined whether there are four or more measurement histories. If the determination in step S142 is NO, in step S143, the latest history is set as a predicted value 30 seconds later, and the flow advances to step S145. If the determination in step S144 is YES, in step S144, a predicted value 30 seconds later is derived from the latest four histories by least squares approximation, and the process proceeds to step S145. . This derivation method is as described in Fig. 18. In step S145, a predicted value after 30 seconds is set. In step S146, the latest history is set to the current value, and the process returns to the processing in FIGS. FIG. 28 is a detailed flowchart of the Web processing capability addition process in step S112 of FIG.

In the flow of FIG. 28, when a coordination center is added, the processing from step S154 is performed.

First, in step S150, the current assigned value is subtracted from the predicted value to determine an additional processing capacity. In step S151, it is determined whether or not there is a spare server in the center. If the determination in step S151 is NO, the process proceeds to step S154. If the judgment in step S15 is YES, step S15 In step 2, an additional server in the center is selected. The details of this processing are as shown in FIG. Then, in step S153, it is determined whether or not the additional processing capacity has been satisfied. If the determination in step S 153 is NO, the process proceeds to step S 154. If the determination in step S155 is YES, the process proceeds to step S158.

In step S154, it is determined whether or not there is a partner center having a preliminary processing capability. If the determination in step S154 is YES, in step S156, the cooperation center allocates processing capacity. Details of this processing are as shown in FIG. In step S157, it is determined whether or not the additional processing capacity has been satisfied. If the judgment in step S157 is NO, step S

Return to 1 5 4. If the determination in step S157 is YES, the process proceeds to step S158. If the determination in step S155 is NO, in step S155, the administrator is warned that the additional processing capacity cannot be satisfied, and the process proceeds to step S158.

In step S158, the VLAN is set to include the selected server, and in step S159, the application is set to the selected server. The application settings are as shown in Figure 22. Step S

At 160, it is determined whether or not there is cooperation between the centers. If the result of determination in step S160 is YES, in step S161, the coordination center load distribution ratio is determined and the equipment is set, and in step S166, the own center and the coordination center are Set the communication band and proceed to step S163.

If the determination in step S166 is NO, the process proceeds directly to step S166. In step S163, the load distribution ratio of the own center is determined, the device is set, and the process returns to the process in FIG.

Fig. 29 shows the data of step S117 of Fig. 25 and the data of step S132 of Fig. 26. It is a detailed flow of database processing capacity addition processing.

In step S170, the current assigned value is subtracted from the predicted value to determine an additional processing capacity. In step S171, it is determined whether or not there is a spare server in the center. If the judgment in step S 171 is N において, in step S 177 the possible web capacity is calculated from the current database, and in step S 178 the insufficient web capacity is added by the cooperation center. I do. The process in step S178 is as shown in FIG. Then, the processing returns to the processing of FIG. 25 or FIG.

If the determination in step S 171 is YES, an additional server in the center is selected in step S 172. Then, in a step S173, it is determined whether or not the additional processing capacity is satisfied. If the determination in step S 173 is NO, the process proceeds to step S 177. If the determination in step S 173 is YES, in step S 174, a VLAN is set to include the selected server, and in step S 175, a database is set for the selected server, In 176, the database list of the Web server in the center is updated, and the processing returns to the processing in FIG. 25 or FIG.

FIG. 30 is a flowchart showing details of the process of selecting an additional server common to the web server and the database.

In step S180, it is determined whether there is a required application server. If the determination in step S 180 is YES, in step S 181, it is determined whether or not there is a server capable of satisfying the additional processing capacity by a single server for the required application. If the determination in step S181 is NO, in step S182, a server for the required use and having the highest performance is selected, and the process returns to step S180. If the determination in step S 181 is YES, in step S 183, the server with the lowest performance among the servers that can satisfy the additional processing capacity with one Select and go to step S188.

If the determination in step S180 is N〇, in step S184, it is determined whether there is an available server. If the determination in step S184 is YES, in step S185, it is determined whether or not there is a server that can satisfy the additional processing capacity with one unit. If the determination in step S185 is NO, in step S186, the server with the maximum performance that can be used is selected, and the flow advances to step S184. If the determination in step S185 is YES, in step S187, the server with the lowest performance is selected from the servers capable of satisfying the additional processing capacity with one server, and the process proceeds to step S188. Proceed to. If the judgment in step S188 is N〇, the process directly proceeds to step S188.

In step S188, a list of assigned servers is constructed, and the process returns to FIG. 28 or FIG. 29. Industrial applicability

According to the present invention, the service quality can be achieved by dynamically allocating the server when it becomes necessary without securing and keeping a sufficient spare server for each service and each data center. In addition, even in the case of a small data center, it is possible to guarantee service quality even in the case of sudden load concentration by linking with other data centers. Furthermore, capital investment can be reduced by sharing the spare server, and at the same time, the equipment can be used effectively.

Claims

The scope of the claims

1. A method for load balancing devices equipped with multiple servers that provide services to clients over a network,

Providing a plurality of spare servers without any service set in the initial state to share the load on the server providing the normal service, and increasing the load on the server providing the normal service. Anticipating, setting an application for a service to be provided to the spare server, setting the application as a server for providing the service, and sharing a load with a server that normally provides the service;

A method comprising:

2. When a plurality of the above devices are connected via a network and one device cannot support the load, the other device has a server that is usually used to provide necessary services. The method of claim 1, wherein the method is provided for the one device.

3. The other device has a spare server, and provides the spare server when a server provided for the one device cannot support a load. The method described in.

4. The method according to claim 2, wherein when sharing a load among the plurality of devices, a communication band between the plurality of devices is secured.

5. In the control step, a predetermined time is calculated based on the number of past requests processed by the server. 2. The method according to claim 1, wherein whether or not the spare server is used for providing a service is determined by predicting a magnitude of a later load.

6. The method according to claim 1, wherein when a spare server is used for a specific service, the spare server is used from a spare server suitable for providing the specific service based on hardware characteristics of the spare server. .

7. The method according to claim 1, wherein when a spare server is used for a specific service, the processing capacity to be replenished is preferentially used from a spare server that can be replenished by one unit.

8. The method according to claim 7, wherein the processing capacity to be replenished by one server is preferentially used from the spare server having the lowest performance among the spare servers that can be supplemented.

9. When a call server is used for a specific service, if there is no spare server that can replenish the processing capacity to be replenished by one unit, the spare server with the highest performance is used. Item 1. The method according to item 1.

10. If the load becomes so small that the load can be supported without a spare server, the control step starts the provision of the service from the spare server used to provide the service with the reduced load. 2. The method according to claim 1, wherein the application for the backup is deleted and the use of the spare server is stopped.

11. The method according to claim 10, wherein when the use of the spare server is stopped, the use of the spare server is stopped in consideration of the characteristics of the hardware of the spare server.

1 2. When the use of the spare server is stopped, the use of the spare server with the highest performance is stopped within a range where the remaining server and the spare server can continue to support the load of the specific service. 10. The method according to 10.

1 3. A device with multiple servers that provide services to clients over a network,

A plurality of spare servers that do not have any services set up in the initial state to share the load on the servers that provide normal services,

In anticipation of an increase in the load on the server that provides the normal service, an application for the service to be provided to the spare server is set, and the server for providing the service is set as the server for providing the service. Control means for sharing, an apparatus comprising:

1 4. A method of load balancing devices with multiple servers that provide services to clients over a network,

In the initial state, none of the services are set to share the load on the server that provides the normal service. Anticipating, setting an application for a service to be provided to the spare server, setting the application as a server for providing the service, and sharing a load with a server that normally provides the service;

A program for causing a computer to realize a method comprising: