JP2003256297A - Application gateway configuration method and system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an application gateway configuration method and a system in which availability equal to a cluster configuration can be realized even without adopting the cluster configuration, clustering software or network constitutive equipment are reduced, costs are reduced, an ASP business can be started with a little funds and application services can be easily implemented. <P>SOLUTION: In the system for an application service provider to provide various application services to a user via a communication network such as Internet, an application gateway server is arranged to function as a front end for services to be provided by the application server, a logical pairing configuration is adopted by the pair of relevant application server and relevant application gateway and in the relevant pairing configuration, the relevant application gateway server controls opening/closing of application services as processing on the side of the relevant application server. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and system for enabling an application service provider (ASP) to provide a user with various application services via a communication network such as the Internet.

[0002]

2. Description of the Related Art Conventionally, in order to pursue availability when an application service provider conducts business, a clustering system is generally adopted. In this case, the gateway server (GW server) is often clustered, and the application server (AP server) is often clustered.

[0003]

In such a conventional system configuration method, special software or hardware addition is required to form a cluster configuration.
There is a problem in that the configuration becomes complicated and it takes time to identify and separate the location of the failure cause at the time of failure.

Further, the application service provider needs to design and build a new system including clustering software according to the type of service to be provided each time the business is started up.
Not only was the cost high, but the period for launching a business was also long.

Furthermore, when adding a new service to an existing service providing system, complicated work such as changing software is required, and this can be performed easily and inexpensively. It wasn't something.

An object of the present invention is to provide a method and system for configuring an application gateway which can solve the above-mentioned problems.

[0007]

According to the concept of the present invention,
When various application services are provided by the application service provider, the service provided by the application server is provided to the communication network such as the Internet via the application gateway server and further provided to the user. The server detects the occurrence of a failure in the application server and the recovery from the failure, and automatically opens and closes the service. More specifically, this is realized by forming a logical pairing with a pair of one gateway server and one application server.

In this pairing, the gateway server controls the open / close processing of the application service, which is the processing on the application server side, to ensure the same availability as in the clustering configuration. Even when configuring two or more gateway servers, the service open / close processing information of the application server is exchanged between the gateway servers and controlled to ensure this pairing configuration, thereby ensuring the same availability as the clustering configuration. .

In order to reduce the cost for designing and constructing such a system and shorten the period for starting a business, the application gateway server has a main function and configuration derived from the service form of an application service provider. Is preferably incorporated in advance.

According to one aspect of the present invention, there is provided an application gateway configuration method in a system for an application service provider to provide various application services to a user via a communication network such as the Internet, wherein A certain application gateway server is arranged so as to function as a front end for the service provided by the server, and a logical pairing configuration is taken with the pair of the application server and the application gateway, and in the pairing configuration, An application characterized in that the application gateway server controls an opening / closing process of an application service, which is a process on the application server side. Touei configuration method is provided.

According to one embodiment of the present invention, the application gateway server and / or the application server have a network cluster configuration.

According to another embodiment of the present invention, when there are a plurality of the application gateway servers, service opening / closing processing information of the application servers can be exchanged between the application gateway servers and controlled.

According to yet another embodiment of the present invention,
In the application gateway server, main functions and configurations derived from a service form called an application service provider are incorporated in advance.

According to another aspect of the present invention, for a service provided by an application server in a system for an application service provider to provide various application services to a user via a communication network such as the Internet. An application gateway server arranged to function as a front end of the instruction gateway server, the application gateway server including a command receiving agent function unit, a state transition agent function unit, and a state scanning agent function unit. The reception agent function unit receives an instruction for transitioning the state of the system, transmits the command to the state transition agent function unit, and activates the state scanning agent function unit to detect the current system status. It operates to notify the bis state,
The state transition agent function unit is activated by the command receiving agent function unit, operates to change the state of the service of the system, and record the transition state and history, and the state scanning agent function unit is configured to operate the application. If the status of the server and another application gateway server exist, the status of the other application gateway server is monitored, the next status is determined according to the detected status, and a command to the command receiving agent function unit is issued. A system is provided which is characterized in that it operates to output

According to one embodiment of the present invention, as an interface for transitioning the system state, "service provision", "service blockage", "service suspension", "service provision state" One set is prepared for one service, with a set of commands such as "hold", "hold service blocked state", "hold service stopped state", and "release service state hold".

According to another embodiment of the present invention, the state transition agent function unit operates according to a state transition program.

According to yet another embodiment of the present invention,
The state scanning agent function unit determines the state to transit to next according to the transition table.

According to yet another embodiment of the present invention,
The state scanning agent function unit is scheduled to perform state scanning at predetermined time intervals that differ according to the state of the system.

According to still another aspect of the present invention, there is provided a computer-readable recording medium in which a program for causing a computer to function as the system described above is recorded.

According to still another aspect of the present invention, there is provided a program for causing a computer to function as the system described above.

[0021]

BEST MODE FOR CARRYING OUT THE INVENTION Next, the present invention will be described in more detail with reference to the accompanying drawings with regard to the embodiments and examples of the present invention.

In order to make the present invention easier to understand, the following explanation will be made by dividing it into items. In the section, the outline of the present invention is described in 2. In Section 3, the basic configuration of the present invention is described in 3.
Section 4 describes the need for standardization of GW servers. And 5. In the section, 6. What is a network cluster and service monitoring that is indispensable for automating the opening / closing process? In the section, regarding opening / closing processing depending on the difference in network configuration, 7. In the section, the elements required to actually perform the opening and closing process and the mounting method are explained.

1. SUMMARY OF THE INVENTION In a system according to the present invention, a service provided by an application server (AP server) is disclosed to the Internet via an application gateway server (GW server). The published URL is 24 hours,
There is a possibility of being constantly accessed for 365 days, and even if the application service is stopped, a mechanism to display that effect (blocking screen) is required. That is, G
The W server displays a blocking screen at the time of maintenance of the AP server and at the time of occurrence of a failure, and restarts service provision at the time of completion of maintenance and recovery from the failure. The GW server is the AP under the environment of the network cluster.
Automates the blocking process when a failure occurs by monitoring the service of the server. Also, when the failure is restored, the service start processing is automatically performed.

When the failure of the AP server is detected and the opening / closing processing of the service is automatically performed, it is necessary to consider the case where the clustering software related to the network is used and the case where it is not. When using the clustering software, the clustering software detects a failure and the network clustering software determines whether or not the service is provided. On the other hand, when the clustering software related to the network is not used, the gateway software (group) on the GW server, which is installed alone or plurally, determines the service providing operation. Here, an automatic opening / closing method applicable to either case will be described.

Before describing the present invention in detail, the meanings of the terms used in this specification will be summarized. 1) "Close" of service This indicates that the service is closed by the GW server providing a closed screen in response to the AP server failure. 2) Service "shutdown" This means that the GW server itself does not provide the service as the front end due to some factor. 3) "Provision (open)" of services Providing services normally. Alternatively, it means that the service is normally provided. 4) “Start” of service This means that the GW server itself starts providing service as a front end.

2. Basic Configuration of the Invention As schematically shown in FIG. 1 of the accompanying drawings, the network topology of the system according to the present invention is logically the Internet 1, GW layer 2, AP layer 3, database (DB).
It is divided into a layer 4 and a storage (SAN / NAS) layer 5.

In the system according to the present invention, the GW layer 2 literally functions as an application gateway. The service provided by the AP layer 3 is provided to the user via the GW 2. The service provided is generally HTTP.
Is a Web-based service using HTTPS and HTTPS, and provides an application service as a whole system.

The AP layer 3 is constructed so as to be protected by the GW layer 2. The security level at the AP layer 3 is the AP layer 3
This is because it is absorbed by the GW layer 2 though it varies depending on the operator of.

From the perspective of security, GW
The server 2 is a firewall, and in terms of functionality, the GW server 2 is a front end for the services provided by the AP server 4.

3. Application gateway server
Standardization of the GW server 2 of the system according to the present invention is configured by one unit or a plurality of units according to a request of a business operator (user) who uses this system. Moreover, the configuration of the AP server 3 is forced to take various configurations depending on the form of the service provided by the business operator.

For each configuration, it is necessary to study the configuration and settings of the GW one by one, and to define the requirements, design, and construct each time. But with this method,
Not only is it costly to start up, but it also takes a long time to start up.

Therefore, the functions for the main items derived from the service form of the application service provider (hereinafter referred to as ASP) are installed in the GW server in advance. That is, by standardizing the configuration of the GW server, the cost required for design and construction is reduced and the period is shortened.

The system as a whole has the following configuration to enhance the availability of the system. When the system is started up, only one AP server, which is a pairing partner corresponding to one GW server, is logically fixed in advance in the setting file on the setting table of the pairing configuration. -The GW server monitors the operating state of its own AP server and reflects it in the service open / close processing status on the GW server. -In the GW server configuration that consists of multiple GW servers,
It has an independent system internal monitoring separate from the cluster software, and exchanges the service open / close status of the GW server with each other by SNMP, grasps the service open / close of the entire system and maintains the consistency.

Items to be examined in order to standardize the blocking process are as follows. 1) Compatible with GW with network cluster (symmetric type) configuration that balances network load 2) Active-Standby (asymmetric)
Corresponding to the GW with a type configuration 3) Only one AP server, or when the AP server can be treated logically as one in the network cluster configuration 4) When there are multiple AP servers and no network cluster configuration 5) Compatible with the case where there is only one GW server and one AP server.
Paying attention to the logical number of servers and AP servers, 1 × 1 configuration (see FIG. 2), 2 × 1 configuration (see FIG. 3), N × 1 configuration (see FIG. 4), 2 × 2 configuration (see FIG. 5). (See FIG. 6) and the N × N configuration (see FIG. 6).

Further, there is a case where the blocking process is automatically performed,
It may be performed manually (when instructed from outside). When a failure occurs in the AP layer, the GW layer detects the failure in the AP layer and automatically performs the blocking process. Further, when the service provision in the AP layer is stopped due to maintenance or the like, the block is manually (from the outside) blocked. In order to provide an environment in which these are integrated, the standardization of the blocking process must take into account the respective configurations.

Since it is necessary to perform the start processing after confirming that all the failures have been recovered, it is generally desirable to perform it manually. However, in the case where the failure itself is automatically recovered, it may be better to perform the start processing automatically.

When the start process is automatically performed, the process shown in FIGS.
In the structure shown in, each must be considered and standardized so as to be consistent as a whole. If the GW server performs the start process without cooperation in the state where the failure is not completely recovered, a “fluttering” phenomenon occurs in which the start and block processes are repeated in a short time, and May also appear to be of poor quality.

As described above, when the opening / closing process is automated, standardization is performed so as to match the whole in each case.

4. Network Cluster Network clusters provide high availability in service provision. By providing a service to one IP address called a service IP address by a plurality of nodes, even if a single node fails, other nodes can continue to provide the service.

There are a symmetric type and an asymmetric type depending on the difference in the cluster configuration method. The former provides the same service with a plurality of nodes providing the service, and the latter switches the node providing the service when a failure occurs. Take the way.

The terms symmetric cluster and asymmetric cluster are called from the viewpoint of the configuration method, and from the viewpoint of function, they are called a scalability cluster and a failover cluster, respectively. For networks,
The symmetric (scalability) type is a load balancing network cluster, and the asymmetric (failover) type is an active standby network cluster. In addition to the active standby type, the standby (Standby)
Cold standby (Cold-S
There are tandby type and hot-standby type.

When a network cluster configuration is used for the AP server, no matter what type of network cluster configuration, a logically single A is seen from the GW server.
It can be handled as a P server. However, when configuring a network cluster with GW servers, it is necessary to consider automation of opening / closing processing.

When considering the automation of the opening / closing process in the network cluster configuration, the following four points must be considered. 1) State of AP server 2) State of other GW servers that make up the cluster 3) Startup of GW server 4) Manual forcible saving of transition state In the concept of high availability, basically two failures occur simultaneously. When it occurs in more than one place, it is generally considered that high availability is not established. However, in the automatic opening / closing process, A
It is considered that high availability is maintained independently in the GW layer even when a failure occurs in the P server. That is,
For the GW layer, a failure of the AP server is a form of one state transition, and it is considered that one failure occurs only when the failure occurs in the GW layer.

4.1. Load balancing network cluster
GW server having a configuration In the case of a GW server forming a load balancing type network cluster configuration, a plurality of GW servers provide the same service.

When the failure of the AP server is detected, it is necessary to synchronize all the GW servers forming the cluster to perform the service blocking process. Further, when the AP server recovers from the failure, all GW servers similarly start the service in synchronization.

When a GW server fails, another GW
There is no need to change the state on the server. This is a major difference from the active standby network cluster configuration.

At the time of starting the GW server, it is necessary to scan the status of the AP server and know the status of other GW servers before starting the service. At the time of starting the GW server, if there is a failure in the AP server, the service is blocked. When the AP server is in a normal state, it makes an inquiry to another GW server and determines its own state.

Further, there may be a case where another GW server is not running. In this case,
6. This will be explained in the section.

4.2. Active standby network
In the case of a GW server with an active standby network cluster configuration, fault detection and automatic blockage processing of the AP server are considered to be equivalent to the case where there is only one GW server.
In other words, when a failure occurs in the Active GW server, the service moves to the Standby GW server side, so that the cluster can be entrusted with the service transfer. However, it is necessary to consider the method of transmitting the open / closed state.

One method is, when the open / closed state changes,
At the same time, the standby GW server should be notified. With this method, the standby machine can immediately take over the state and start the service. However, when the standby GW server is a cold standby (the system according to the embodiment of the present invention does not have a cold standby GW server), a failure is recovered, or a planned restart is performed. It is possible that the status has not been notified.

The other method is to move the service to the other side, and the AP that is newly activated on the GW server side.
Checking the service status of the server. In this method, the active GW server can always take the initiative in ascertaining the state. However, there are disadvantages that it takes time to check the service status after the service is taken over by the cluster and that the status cannot be taken over when the blocking process is manually performed.

As a compromise between these two methods, the service status of the AP server is investigated when the service is started up,
If it is known that there is a clear failure, put it in the closed state. If the service status of the AP server is normal, make an inquiry to the active GW server,
Record the condition.

Furthermore, there is a case in which, when the active GW server is started up, no GW server other than the GW server is started up. In this case, since it is the same as the case where the number of GW servers that do not form a cluster is one,
6. This is explained in Section.

5. Server for application server
Screw monitoring In order to perform automatic opening / closing processing, it is essential to monitor the service provided by the AP server. For example, A
When the P server provides the HTTP Web service of TCP port 80, the GW server uses the port 8
HTTP access is attempted for No. 0 and the service of the AP server is monitored. The expected response should be returned in the normal provision state, and if an error or a connection error occurs at the protocol level, it is determined that there is a failure in the AP server service or the network for reaching the AP server. .

Since the GW server itself sets a time interval and performs automatic monitoring, when there are a plurality of GW servers, the respective GW servers have different timings for scanning the service of the AP server. Already 4. As described in the section, due to this difference in timing, the state of the AP server detected by the GW server group may be different. FIG. 7 schematically illustrates the relationship between such monitoring timing and the detected state.

If the detected states are different, the open / closed state will be inconsistent when viewed as a whole GW server group. That is, the GW server that has detected the failure of the AP server is in the service blocking state, and the GW server that has not yet detected the failure of the AP server is in the service providing state.
In the case of a load-balanced network cluster configuration, the service may be blocked or provided by the GW server to be distributed (actually, the AP server has a failure, so the failure can be detected. Error occurs on GW servers that are not.)

In order to prevent such a situation, the GW server that first detects the failure of the AP server must notify the detected failure information to other GW servers by broadcast communication.

6. Automatic opening due to differences in network configuration
Regarding the necessity of standardizing by absorbing the difference in the configuration of the closed processing network, see 3. As described in Section, this section describes specific details to be considered. When there are a plurality of GW servers, it is assumed that they all have a load balancing network cluster configuration. In addition, we do not consider two-point disorder. The failure of the AP server is a state transition for the GW server and is not regarded as the first failure in high availability. When a failure occurs in the GW server group and the network connecting them, it is considered as the first failure for the first time.
For the status related to the service of the GW server, "provision",
There are three types: "block" and "stop".

6.1.1 × 1 Configuration As shown in FIG. 2, this is the simplest configuration and is the basic configuration of all. The configuration is such that there is one GW server and one AP server. The same applies when the AP server forms a network cluster configuration and logically appears as one unit.

Items necessary for the GW server in this configuration are as follows. 1) When a failure of the AP server is detected by monitoring, the service is blocked. 2) When the return of the AP server is detected by monitoring, the service is started (provided). 3) If a manual block command is issued, block the service. At this time, the status monitoring of the AP server is stopped. That is, the service is not started unless there is a manual command again. 4) When a manual start (provide) command is issued, the service is started and monitoring is restarted. 5) If the blocking of the service and the start of the service are repeated the set number of times within a fixed time, it is determined that another failure has occurred, and the service is blocked. 6) Check the status of the AP server at startup, and shut down the service if a failure has occurred. If no failure has occurred, the saved flag is confirmed, and if a blocking command from the outside is issued, the service is blocked. When the AP server operates normally and there is no block command from the outside, the service is restarted.

6.2.2 × 1 Configuration As shown in FIG. 3, the number of AP servers is logically one,
When two GW servers are configured, two GW servers need to cooperate in addition to the 1 × 1 configuration case. The two GW servers each monitor the service of the AP server and perform the blocking process as in the 1 × 1 configuration. At this time, depending on the monitoring timing (time interval), a situation occurs in which one side blocks the service and the other side provides the service. In other words, the whole system is not consistent. In order to eliminate this inconsistency, when one GW server detects a failure, the other GW server is instructed to check the state. GW instructed to check the status
On the server side, the scheduled status confirmation schedule is reset and scanning is immediately performed. The operation for this purpose will be described below in further terms.

6.2.1. Detection and operation in the provision state 1) When the AP server is normal and the other GW server is in the provision state, the provision state is maintained. 2) When the AP server is normal and the other GW server is in the blocked state, the provision state is maintained and the other GW is
Tell the server to check the status. 3) If the AP server is normal and the other GW server has stopped the service or a failure has occurred, maintain the provision state and instruct the other GW server to check the state. To do. 4) If the AP server fails and the other GW server is in the provision state, stop the service and
Instruct the W server to check the status. 5) If a failure occurs in the AP server and the other GW server is in a blocked state, block the service. 6) If a failure occurs in the AP server and the other GW server stops the service or the failure occurs, the service is blocked.

6.2.2. Detection and operation in closed state The same operation as detection and operation in provided state is performed. 1) When the AP server is normal and the other GW server is in the provision state, the service is provided. 2) If the AP server is normal and the other GW server is in the blocked state, provide the service and then the other GW
Tell the server to check the status. 3) If the AP server is normal and the other GW server stops the service or has a failure, provide it and instruct the other GW server to check the status. 4) If the AP server fails and the other GW server is in the provision state, stop the service and
Instruct the W server to check the status. 5) If a failure occurs in the AP server and the other GW server is in the blocked state, the blocked state is maintained. 6) When the AP server fails and the other GW server stops the service or the failure occurs, the blocked state is maintained.

6.2.3. Detection and operation in the stopped state 1) If the AP server is normal and the other GW server is in the provision state, bring it into the provision state. 2) When the AP server is normal and the other GW server is in the blocked state, the provision state is set, and the other GW server is instructed to confirm the state. 3) When the AP server is normal and the other server has stopped the service or the failure has occurred, the AP server is brought into the provision state, and the other GW server is instructed to confirm the state. 4) If a failure occurs in the AP server and the other GW server is in the provision state, the stop state is maintained and the other G server
Instruct the W server to check the status. 5) When a failure occurs in the AP server and the other GW server is in the blocked state, the stopped state is once maintained and the transition to the blocked state is tried. 6) When a failure occurs in the AP server and the other GW server stops the service or the failure occurs, the stopped state is once maintained, an attempt is made to shift to the blocked state, and another Instruct the GW server to check the status.

G in the pairing configuration as described above
The operation of the AP server opening / closing process between the W servers is summarized as shown in Table 1 shown in FIG. Furthermore, FIG.
Shows, in the form of a flow chart, the operation when the state of the own GW server is the providing state during the operation shown in Table 1, and corresponds to the operation shown in the upper part of Table 1 in FIG. FIG. 10 shows an operation in the form of a flow chart when the state of the own GW server is the blocked state, and corresponds to the operation shown in the middle part of Table 1 of FIG.
FIG. 11 shows an operation in the form of a flowchart when the state of the own GW server is in the stopped state, and Table 1 of FIG.
This corresponds to the operation shown in the lower part of FIG.

6.3. N × 1 configuration The configuration is schematically shown in FIG. 4, and synchronization is established between the GW servers as in the case of the 2 × 1 configuration. When notifying other GW servers, all the GW servers are notified by using broadcast communication.

Although it is almost the same as the case of the 2 × 1 configuration, the G confirming the status is notified by the broadcast communication and the G is received.
On the W server side, after operating the status, a reply is returned. The reply is also broadcast so that all other GW servers can update the status.

Under the following conditions, it is necessary to replace Table 1 of FIG. 8 shown in the 2 × 1 configuration. 1) Of the GW servers that received a reply within a certain period of time,
If there is at least one GW server in the provision state, another GW
Bring the server into the provision state. 2) Of the GW servers that received a reply within a certain period of time,
If there is no GW server in the provision state and at least one GW server is in the blocked state, the other GW server is set in the blocked state. 3) Of the GW servers that received a reply within a certain period of time,
If there is no GW server in the provided state and no GW server in the blocked state exists, the other GW server is set as a failure or a stop. 4) If there is no reply within a certain period of time, the other GW server is set to a failure and stopped.

The operation described in Table 1 of FIG. 8 is once performed, the reply returned within a fixed time is confirmed, and the state is decided again. When the operation is determined by the reply so that the loop does not occur, the status confirmation instruction to other GW servers is not issued.

6.4.2 × 2 Configuration This is the configuration schematically shown in FIG. 5, in which the 1 × 1 configuration is duplicated. In this case, even if one GW server detects a failure of the AP server, the other GW server
You can continue to provide services. Actually 2x
In the case of 2 configurations, the state detection and the corresponding operation are 2 ×
This is the same as in the case of one configuration. That is, the operation is determined based on Table 1 in FIG.

6.5. N × N configuration The configuration is schematically shown in FIG. 6, and is similar to the case of the N × 1 configuration. That is, 6.3. It may be considered in the same manner as the operation described in the section. 6.6. For standardization , all cases can be covered using the basic operation of 1 × 1 configuration and the operation of N × 1 configuration.

State detection and state transition are closely related as shown in Table 1 of FIG. 8, but by separating each method, various methods can be applied. In particular, if you want to block the AP server for maintenance, etc., by making the state transition method independent and preparing an interface from the outside,
It is possible to use.

The method for detecting the status is to scan the service status of the AP server and scan the status of other GW servers.

The state transition consists of three provisions of service, blockage, and suspension, and interfaces for external instructions are prepared. When used for maintenance, regardless of the status of the AP server or other GW server,
You may want to keep the same state. For this reason, the provided state is maintained (Keep open) and the blocked state is maintained (Keep clos
e), three states of keep state (Keep shutdown),
Prepare an interface for instructing each from the outside. In addition, it is necessary to prepare an interface for instructing the cancellation of state retention (No Keep).

7. Implementation Method Next, a specific method and configuration for implementing the method described in the previous section will be described by further dividing them into terms. 7.1. State transition method 7.1.1. State Transition Interface In a system as an embodiment of the present invention, an SNMP trap (instruction) is used as an interface for changing states. The SNMP trap is set to be accepted only from the own GW server and the server on the local network.

6.6. As described in the section, prepare the following seven interfaces for one service. 1) open: service provision (start) 2) close: service closure 3) shutdown: service stop 4) Keep open: service provision status retention 5) Keep close: service closure status retention 6) Keep shutdown : Keeping the stopped state of the service 7) No keep: Releasing the maintenance of the state of the service Since these interface sets are prepared for each service, if there are M services provided by the GW server, (M × 7) Will have individual interfaces.

7.1.2. Program for state transition A program that starts up after receiving a trap audits the sender of the trap and determines whether it can operate. The state transition program simply performs the instructed operation, and operates separately from the relationship with other factors. 7.2. Status scanning method The status of each GW server is scanned by making a status inquiry by SNMP.

With respect to the AP server in charge, it suffices to simply scan whether the service is provided. However, the GW server may not be simply scanned when using a network cluster.
This is because there is a case where a virtual address is assigned by the cluster and the service on the GW is normally provided only when the virtual address is accessed.

This corresponds to the case where the virtual address of the cluster and the virtual IP host of Apache are used. In such a case, even if an attempt is made to confirm the service by using the real address of the GW server to be scanned, the obtained result is different from the actual service provision state (the same service is used for the virtual address and the real address. The same result can be obtained only when provided). Further, even if the virtual address for the service is used, it is not always possible to scan the state of the target GW server.
Therefore, it is better to think that there is no method to inspect the service provided by a specific GW server from the outside (it is possible by modifying the clustering software, but it depends on the structure of the clustering software, and in open source, If you use software that does not exist, it may not always be possible to modify it, so using this method may not always be practical).

Therefore, the service status of each GW server preferably takes the form of an inquiry to each GW server and a self-reported reply to the inquiry, instead of scanning the service. Therefore, SNMP is used to make an inquiry to each GW server.

However, "holding the provision state" of the service,
If the service is in the "hold status" or the service is in the "hold status", scanning is not performed and the specified status is held.

If the result of the inquiry is any one of the state retention, the state of the GW server is considered to be the state of no response. This is because it is considered that the state retention is due to an instruction from the outside, and that the inconsistent state due to the state should be resolved by the responsibility of the person (person, program) who gives the instruction from the outside. is there.

7.3. Status Notification In the system as one embodiment of the present invention, UCD-SNMP (NET-SNM) is used as a command receiving agent.
P) is adopted, and the response to the inquiry can be easily implemented. Software for providing a service or a wrapper (Wrapper) for the software is prepared, and these describe and hold the current state in a file. For the response to the status inquiry, refer to this file and reply.

7.4. Determination of Next Transition State It has a transition table (see Table 1) to the next state according to the current state and other factors (the state of the AP server, the state of the other GW server, and the state to be held). According to this table, the next transitional state is determined. 7.5. Scheduling of state scanning In the provision state, state scanning is performed using a predetermined time interval (about 10 minutes). In the closed and stopped state, the state scan is performed at 1-minute intervals. This is to respond quickly to disaster recovery. In the case of state holding, the state scanning scheduling is stopped until the state holding is released.

7.6. State transition history Record the state transition history as a log in a file. This is to detect a case where the state changes significantly in a certain period of time. There may be a failure that cannot be handled by automatic failure detection, such as when the network on the AP server side is unstable. For example, in the case where provision / blocking is repeated 10 times in 5 minutes, the service should be blocked or stopped (see 6.1). 7.7. Status scanning instructions and responses Prepare an interface for forcibly performing status scanning. 7.3. Similar to the item above, it is provided by an SNMP trap. The command reception agent of the GW server that is requested to scan the state scans only the state of the AP server, unlike the normal state scan, and transits the state assuming that the other GW server has stopped the service. Let The result of the state transition is returned to all GW servers by an instruction (SNMP trap).

7.8. Specific Configuration of System The specific configuration of the system as described above is summarized as shown in Table 2 of FIG. 12, and FIG. 13 shows this in a block diagram. FIG. 13 shows the above 6.4. 2 illustrates an example of the configuration of the application gateway server in the 2 × 2 configuration system described in the section and the mutual relation.
As shown in FIG. 13, one GW server 2A
Is paired with one AP server 3A, and the other GW server 2B is paired with the other AP server 3B. Each GW server includes a command reception agent function unit 11, a state transition agent function unit 12, and a state scanning agent function unit 13. As described above, the command receiving agent function unit 11 receives the command for changing the state of the system, transmits the command to the state transition agent function unit 12, and activates the state scanning agent function unit 13 to activate the system. Operates to notify the current service status of the. As described above, the state transition agent function unit 12 is activated by the command reception agent function unit 11 and operates to change the service state of the system and record the transition state and history. As described above, the state scanning agent function unit 13 uses the AP server 3A.
Alternatively, the state of 3B and the state of the GW server 2B or 2A are monitored, the next state is determined according to the detected state, and the command receiving agent function unit 11 is operated so as to issue a command.

Further, the state transition agent function unit 12
Operates according to the state transition program as described above. Furthermore, the state scanning agent function unit 1
3 determines the next transition state according to the transition table as described above. Then, as described above, the state scanning agent function unit 13 is scheduled to perform the state scanning at predetermined time intervals which are different depending on the state of the system.

The opening / closing process of the application service in the system according to the embodiment of the present invention has been described. When the AP server itself detects a failure, the case where the blocking process is manually performed forcibly is performed. Since it is similar, it is not particularly described. Furthermore, the present invention can be extended and applied even when it is blocked depending on the time of day.

[0089]

According to the present invention, even when a plurality of GW servers are configured, by taking a pairing configuration,
Even if the cluster configuration is not adopted, availability equivalent to that of the cluster configuration can be realized.

Further, since it is not necessary to have a plurality of network devices such as HUB and LAN placed between the GW server and AP server, clustering software,
There are fewer network components and the cost is lower.

Furthermore, incidentally, since the cost for the ASP business to start up the business is released from the cluster configuration taken from the beginning, the ASP business can be started with a small amount of capital, and the initial cost for the user can be reduced. Since the burden of service provision cost can be reduced, it becomes easy to implement the application service.

[Brief description of drawings]

FIG. 1 is a diagram for explaining a network topology of a system according to the present invention.

FIG. 2 shows a GW server and A in the system according to the present invention.
It is a figure which outlines the 1x1 structure of a P server.

FIG. 3 is a GW server and A in the system according to the present invention.
It is a figure which shows roughly the 2x1 structure of a P server.

FIG. 4 is a GW server and A in the system according to the present invention.
It is a figure which shows roughly the Nx1 structure of a P server.

FIG. 5: GW server and A in the system according to the present invention
It is a figure which outlines the 2x2 structure of a P server.

FIG. 6 is a GW server and A in the system according to the present invention.
It is a figure which shows roughly the NxN structure of a P server.

FIG. 7 is a diagram illustrating a relationship between a monitoring timing and a detected state in the system according to the present invention.

FIG. 8 is a diagram showing Table 1 summarizing the operation of AP server opening / closing processing between GW servers in the pairing configuration of the system according to the present invention.

9 is a diagram showing the operation in the form of a flowchart when the state of the own GW server is the providing state during the operation shown in Table 1 of FIG.

10 is a diagram showing the operation in the form of a flowchart when the state of the own GW server is the blocked state during the operation shown in Table 1 of FIG.

11 is a diagram showing an operation in the form of a flowchart when the state of the own GW server is the stopped state during the operation shown in Table 1 of FIG.

FIG. 12 is a diagram showing Table 2 in which a specific configuration of the system according to the present invention is summarized.

13 is a diagram showing a specific configuration summarized in Table 2 of FIG. 12 in the form of a block diagram.

[Explanation of symbols]

1 Internet 2 GW layer 2A GW server 2B GW server 3 AP layer 3A AP server 3B AP server 4 DB layer 5 storage layers 11 Command receiving agent function unit 12 State transition agent function 13 State Scan Agent Function

Claims (11)

[Claims] 1. A method for configuring an application gateway in a system for an application service provider to provide various application services to a user via a communication network such as the Internet, which is for a service provided by an application server. A certain application gateway server is arranged so as to function as a front end, and a logical pairing configuration is formed by a pair of the application server and the application gateway. In the pairing configuration, the application gateway server is the application server. A method for configuring an application gateway, characterized in that opening / closing processing of an application service, which is a side processing, is controlled. 2. The application gateway configuration method according to claim 1, wherein the application gateway server and / or the application server has a network cluster configuration. 3. The service opening / closing processing information of the application server can be exchanged and controlled between the application gateway servers when there are a plurality of the application gateway servers.
The application gateway configuration method described in. 4. The application gateway configuration method according to claim 1, 2 or 3, wherein the application gateway server has a main function or configuration derived from a service form called an application service provider, incorporated in advance. 5. A system for an application service provider to provide various application services to a user via a communication network such as the Internet, so that the application service provider functions as a front end for services provided by an application server. The application gateway server is provided, and the application gateway server includes a command receiving agent function unit, a state transition agent function unit, and a state scanning agent function unit,
The command receiving agent function unit receives a command for transitioning the state of the system, transmits the command to the state transition agent function unit, and activates the state scanning agent function unit to determine the current service state of the system. The state transition agent function unit is activated by the command reception agent function unit to change the state of the service of the system, record the transition state and history, The scanning agent function unit monitors the state of the application server and the state of the other application gateway server, if any, and determines the next state according to the detected state. Designed to issue a command to the command receiving agent function unit System which is characterized in that. 6. An interface for transitioning a system state, which includes "providing a service", "blocking a service", "stopping a service", "holding a provided state of service", and "holding a blocked state of service" 1 for one service, with a set of commands such as "," holding the stopped state of the service ", and" releasing holding the state of the service "
The system according to claim 5, which is provided as a set. 7. The system according to claim 5, wherein the state transition agent function unit operates according to a state transition program. 8. The system according to claim 5, 6 or 7, wherein said state scanning agent function unit determines a state to transit to next according to a transition table. 9. The state scanning agent function unit according to claim 5, wherein the state scanning agent function unit is scheduled to perform state scanning at a predetermined time interval which is different depending on the state of the system. system. 10. A computer-readable recording medium recording a program for causing a computer to function as the system according to any one of claims 5 to 9. 11. A program for causing a computer to function as the system according to any one of claims 5 to 9.