JPH0863362A - Managing function imparting method and its platform - Google Patents

Abstract

PURPOSE: To increase the efficiency of system operation, to decentralize an excessive load, and to improve the reliability. CONSTITUTION: Each unit platform confirms the priority of an application program to be processed, a module in use, the mean value of times regarding the operation, an operation process history, etc., by a fault diagnostic means 5 and an operation state diagnostic means 7 to mutually inform other platforms on the network, and each unit platform compares reported information with its own load information and gives a specific unit platform the management right according to predetermined algorithm. The unit platform given the management right performs the load management of all unit platforms, selects respective unit platforms in light load states by a selecting means 10 according to information from the respective unit platforms, and switches connections by a connection switching means 11 to make requests for operation processing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a management function in a distributed management function agreement type platform for use in combining a database group and an application program group in the management and operation of networks used in the communication field, office field, etc. The present invention relates to a granting method and its platform.

[0002]

2. Description of the Related Art A conventional technique will be described with reference to FIG. FIG. 2 shows a database DBA in which application programs AP1 and AP2 store data groups 1 and 2, respectively.
, DBB are used to perform a series of jobs. As shown in the figure, a database (hereinafter, abbreviated as DB) and an application program (a program that performs a series of jobs using the corresponding DB). Hereinafter, it is directly referred to as AP). Therefore, it is difficult to distribute the data in the DB, and when the DB is changed, it is necessary to change the AP. That is, AP1 cannot perform a series of jobs using both the data group 1 of DBA and the data group 2 of DBB, and it is necessary to newly create an AP or system.

Further, as shown in FIG. 3, even in a module that performs a specific job, the AP and the module M are directly connected. Therefore, if the module M is changed, AP
Had to change too. Furthermore, a plurality of modules M
The change in the case of an AP using multiple DBs was very complicated.

As a means for solving such a problem, as shown in FIG. 4, AP and DB, AP and module M are separated without being directly connected, and a platform (hereinafter abbreviated as PF) which is a connecting portion for connecting them. ), The distribution of data is facilitated, and the DB and module M
A method has been proposed to mitigate the impact of changes in AP on APs.

Here, PF is a plurality of DBs and modules M
Is software for connecting the AP and
IEICE VOL.E76-B, No.4, April 1993, pp.
391-401 N. Kashima et.al. "Optical cable network op
eration in subscriber loops ".

However, as shown in FIG. 4, in the situation where a large number of DBs, modules M, and APs are connected via the PF, there is a risk that the APs will be concentrated on the PF. Therefore, if the waiting time of work is long or the load is not distributed among a plurality of PFs, in the worst case, there is a possibility that the PF may fail due to excessive load, resulting in an inefficient system configuration. It was In addition, the failure of the PF has a great influence on the entire system. That is, the failure of the DB, the module M, and the AP is local, but the failure of the PF is all DBs and modules M connected to the PF.
The AP may be affected, and the system may be unreliable.

Further, in general, as a means for improving reliability, an N-fold redundant system for connecting a plurality of identical systems in parallel is adopted. However, even if no load balancing means is applied here, the load concentrates only on a specific PF, which reduces the processing speed of the entire system, induces a failure due to excessive load, and causes a light load state (free state). Including the existence of PF).

For this load management control, software provided with a centralized management control means called a system switcher as shown in FIG. 5 is referred to as, for example, "subscriber line OpS".
Highly reliable platform configuration ", IEICE Spring Conference B-
990 (1994). The system switcher recognizes the load status of each unit PF from the self-operation status information from each unit PF, and distributes the load so as to maximize the processing efficiency in the entire system. Further, when it is determined from the self-failure information that one of the units PF has failed, the unit PF with the lightest load among the other units PF on the network is avoided in order to avoid delay in the processing of the work requested by the unit PF. The feature is that you can select to continue the work.

[0009]

However, regarding software having a centralized management control means called a system switcher, when considering the reliability of the entire system, when the centralized management control function of the system switcher itself fails, The load concentrates only on the unit PF, and affects the entire system. Also, the failed unit P
Continuation of the stagnant work that F was processing is not executed efficiently, and therefore the waiting time of the operator may be extended.

In view of the above problems, an object of the present invention is to improve the efficiency of system operation and to distribute an excessive load,
Another object of the present invention is to provide a method of providing a management function of a distributed management function agreement type platform that can improve the reliability of the management control of the load distribution and the platform.

[0011]

In order to solve the above problems, the present invention provides, in claim 1, a database group, a task module group for performing a specific task, and an application program for performing a series of tasks using the two groups. Group, 3 above
For a platform in a distributed processing system in which a network is constructed using a plurality of unit platforms that are the connection part with a group, each unit platform is related to the priority of the application program processed by itself, the module being used, and the operation Check the operating status and failure status such as the average value of time and business processing history, and notify each other of such information to other unit platforms on the network.Each unit platform notifies the notified information and its own load. The first process of comparing the information and giving the management right to the specific unit platform according to the predetermined algorithm, and the unit platform to which the management right is given,
Performs load management for all unit platforms, selects each unit platform in a light load state based on the information from each unit platform,
The second process of switching connection for requesting business processing,
When the unit platform to which the management right is given falls into an overload state, the management right is abandoned and a third process of selecting a unit platform having a new management right is included.

According to a second aspect of the present invention, a database module, a task module group that performs a specific job, an application program group that performs a series of jobs by using the two modules, and the third module.
In a platform in a distributed processing system in which a network is constructed by using multiple unit platforms that are the connection parts with each group, each unit platform can determine the priority of the application program to be processed by itself, the modules in use, and the operating status. Operation status diagnosis means for confirming the operation status such as the average value of the time concerned, work processing history, etc., operation status notification means for mutually notifying other unit platforms on the network, and own failure Failure diagnosis means for recognizing the situation, failure notification means for notifying the failure status,
The unit platform to which the management right is given manages the load of all unit platforms, and the selection means for selecting each unit platform in the light load state based on the information from each unit platform, and for the business processing request And a connection switching means for switching the connection.

According to a third aspect of the present invention, in the platform according to the second aspect, the unit platform to which the management right is given has another unit platform failure diagnosis command transmission means which is a diagnostic function for grasping the failure status of all unit platforms. I had it.

[0014]

According to the first and second aspects of the present invention, each unit platform has a priority of an application program to be processed by itself, a module in use, an average value of operating time, a business processing history, etc. The operating status and failure status are confirmed by the operating status diagnostic means and the failure diagnostic means, and the information is mutually notified to other unit platforms on the network by the operational status reporting means and the failure reporting means, and each unit platform is notified. The obtained information is compared with its own load information, and a management right is given to a specific unit platform according to a predetermined algorithm. The unit platform to which the management right is given manages the load of all unit platforms, selects each unit platform in the light load state by the selection means based on the information from each unit platform, and performs the work by the connection switching unit. Switch connection for processing request. When the unit platform to which the management right is given falls into the overload state, the management right is abandoned and the unit platform having the new management right is selected in the same manner as described above.

According to the third aspect, when each unit platform has a hardware problem and cannot provide any information to the unit platform to which the management right is given, it is externally determined whether the unit platform is actually broken. to decide.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to FIGS.
0 will be described.

FIG. 1 shows a schematic configuration of each unit PF as software stalled in an electronic computer. As shown in FIG. 6, a plurality of units PF and AP are provided.
Group, M group is DCN (Data Communicatio) like LAN.
n Network) to form one distributed processing system. Each unit PF serves as a processing unit, and is an AP that interfaces with a plurality of AP1 and AP2.
An interfacer 1, an M interfacer 2 that interfaces with a plurality of modules M1 and M2, a DB interfacer 3 that interfaces with DB1
An accessor 4 that manages access requests from a plurality of APs 1,2, modules M1,2, DB1 via the respective interfacers, and a failure diagnosis function unit that diagnoses a failure of the unit PF itself as a unit PF management function unit. 5, a failure notification function unit 6 for notifying the failure of the unit PF itself, an operation status diagnosis function unit 7 for checking the processing status in the unit PF, and an operation status notification function unit 8 for notifying the processing status in the unit PF. Have. Then, the failure information from the failure notification function unit 6 and the operation status information from the operation status notification function unit 8 are supplied to the operation status information storage unit 9 of each unit PF in the system.

The unit PF control function unit in each unit PF has a selection function unit 10, a connection switching function unit 11, and a PF failure diagnosis command transmission function unit 12 of another unit. These are for controlling the load distribution of all units PF in the network.

The selection function section 10 is a failure information and operation status notification function section 8 from the failure notification function section 6 of the unit PF.
The connection switching function unit 11 issues a switching instruction to the DB, the module M, and the AP group based on the operation status information from the. Also,
Regarding the connection, selecting the unit PF that is in the lightest state for load leveling and processing the overloaded work leads to efficiency of processing speed and shortening of operator waiting time. The selection is optimized by the selection function unit 10 in which the unit PF to which the management right is given is incorporated in its own control function unit based on the failure information and the operation status information from all the units PF on the network. Select the appropriate unit PF.

The detection of a failure in each unit PF is regularly performed by using the failure diagnosis function section 5 in the unit PF, and when a failure is detected by this self-failure diagnosis function,
The failure notification function unit 6 of the unit PF is used to notify the unit PF to which the management right is given. The unit PF to which the management right is given executes the system switching so that the AP group is not connected to the unit PF which has self-declared the failure. The operation status of the unit PF is periodically notified to the unit PF to which the management right is given by the operation status notification function unit 8. The operation status information to be reported is the work currently performed by the unit PF, the priority of the AP in the work,
The AP group and modules used for the work, and the operating time required to complete the work. As a result, the unit PF to which the management function is added and each unit PF are
It is possible to know the status of other units PF in the system. The unit PF to which the management right has been given receives information about the presence or absence of a failure from another unit PF and a notification of the operation status, accumulates this in the operation status information storage unit 9, and appropriately stores the information based on this information. Connection function to the unit PF 1
Switching work is performed with 0.

If the computer in which the unit PF is installed has a hardware problem and the failure diagnosis function of each unit PF does not operate, the state of the unit PF cannot be grasped. In this case, the unit PF to which the management right is given transmits a signal to all the units PF by the other unit PF failure diagnosis command transmission function unit 12, and the unit PF receiving the signal is operating normally. A signal notifying that there is is returned to the unit PF to which the management right has been given. Here, the unit PF to which the management right is given is a unit PF that cannot reply to the diagnostic signal.
Is judged as a failure. As a result, even if the electronic computer causes a hardware problem and the failure diagnosis function of each unit PF does not operate, the unit PF that has failed can be identified.

In FIG. 7, the unit PF with the lightest load among the units PF on the network is the unit PF.
The flow of being given the management right of is shown.

FIG. 8 shows a flow in which the management right is abandoned and a new management right is given when the load state of the unit PF to which the management right is given becomes overloaded. Here, the method of selecting the unit PF that entrusts the management right is as shown in FIG.

The unit PF which is not given the management right diagnoses its own failure / operation status and overloads the unit PF.
9 and 10 show an example of a method and a flow in which the unit PF to which the management right is given distributes the load in the case where there is an error. The contents of the operation status are the business process history and the AP to be processed.
Priority, module in use, average time taken for operation, etc. As a result, the unit PF in the light load state is
Can be grasped and the unit PF can be preferentially connected when the next work occurs, so that the load on the unit PF can be dispersed.

[0025]

As described above, according to the present invention, the efficiency of system operation can be improved, an excessive load can be distributed, and the management control function section is held by each unit PF. Since the unit PF, which has the least load for business processing, is responsible for the management control function unit, it is possible to carry the management function unit around in response to the ever-changing load status. As a result, even if the unit PF to which the management right is given has a problem, another unit PF can substitute the failed unit PF. Therefore, the redundancy of the management control function unit provides the reliability of the entire system. There is an advantage that the property can be enhanced.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining the present invention.

FIG. 2 is a diagram illustrating a conventional technique.

FIG. 3 is a diagram illustrating a conventional technique.

FIG. 4 is a diagram illustrating a conventional technique.

FIG. 5 is a diagram illustrating a conventional technique.

FIG. 6 is a configuration diagram for explaining the present invention.

FIG. 7 is a diagram showing an example of a flowchart of the present invention.

FIG. 8 is a diagram showing an example of a flowchart of the present invention.

FIG. 9 is a diagram showing an example of a flowchart of the present invention.

FIG. 10 is a diagram showing an example of a flowchart of the present invention.

[Explanation of symbols]

1 ... AP interfacer, 2 ... Interfacer, 3 ... D
B interfacer, 4 ... accessor, 5 ... failure diagnosis function section, 6 ... failure notification function section, 8 ... operation status diagnosis function section, 9
... Selection function unit, 10 ... Connection switching function unit, 11 ... Other unit PF failure diagnosis command transmission function unit, AP1, AP2, AP3
… Application program, M1, M2, M3… Module, PF1, PF2, PF3… Unit platform.

Claims (3)

[Claims] 1. A network using a plurality of unit platforms which are a group of a database group, a task module group for performing a specific job, a group of application programs for performing a series of tasks using the second group, and a unit section for connecting to the third group. In the distributed distributed processing system platform, each unit platform has its own priority of the application program to be processed, the module being used,
Check the operating status and failure status such as the average value of the time required for operation, business processing history, etc., and notify each other of the information to other unit platforms on the network. The first process of comparing the load information of its own and assigning the management right to a specific unit platform according to a predetermined algorithm, and the unit platform to which the management right is assigned manages the load of all unit platforms. Based on the above information from each unit platform, select each unit platform in a light load state, and perform the second process of switching connections for business processing requests and the unit platform to which the management right has been granted. If you fall into a load condition, give up control of it and manage Management functionalization method characterized by having a third step of selecting a unit platform with. 2. A network using a plurality of unit platforms which are a group of database, a task module group performing a specific job, an application program group performing a series of jobs using the two groups, and a unit connecting with the three groups. In the platform of the distributed processing system that is configured, each unit platform has its own priority of the application program to be processed, the module being used,
An operation status diagnostic means for confirming the operation status such as the average value of the operation time, business processing history, etc., and an operation status reporting means for mutually notifying such information to other unit platforms on the network, The failure diagnosis means for recognizing the failure status of the, the failure notification means for notifying the failure status, and the unit platform to which the management right is given manages the load of all the unit platforms, and from each unit platform 2. A platform comprising: a selection unit that selects each unit platform in a light load state based on the above information, and a connection switching unit that performs connection switching for a business processing request. 3. The unit platform to which the management right has been granted has another unit platform failure diagnosis command transmission means, which is a diagnostic function for grasping the failure status of all unit platforms. platform.