JP2010086317A - Resource allocation method, resource allocation program, and flow processing system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To suppress a decrease in reliability of a flow execution path in allocating each of nodes called from a flow of a flow processing system to a plurality of physical computers. <P>SOLUTION: When an operation management device 1 cannot allocate a service called by a whole BP (Business Process) to one service execution device 3, the device 1 divides the whole BP into a plurality of partial BPs along an execution path for each branch of the whole BP, calculates total necessary resource amount information on the service called by executing the BP to determine a necessary resource amount of the partial BP, retrieves the service execution device 3 which has an available resource amount larger than the determined necessary resource amount of the partial BP, and allocates the service called by partial BP to the retrieved service execution device 3. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a resource allocation method, a resource allocation program, and a flow processing system.

Virtual computer technology is used that logically divides physical computer resources and achieves the same effect as using multiple computers. By using virtual computer technology, the number of physical computers can be aggregated by executing operations that have been executed on separate physical computers with virtual computers on a small number of physical computers. Patent Document 1 discloses a technology for allocating resources of a physical computer according to a load on a virtual computer.
JP 2002-202959 A

  Many business operations executed on a computer are expressed by a flow format in which a plurality of business units are connected according to an execution order. The business flow expressed in the flow format is divided into a plurality of execution paths by branching. In the conventional technology, each business unit is individually assigned to a separate physical computer, but the execution path of the flow is not taken into account, so if the reliability of a part of the execution path decreases, the execution path starts. Decreasing reliability affects the entire route from the point to the end point.

FIG. 20 is an explanatory table showing corresponding terms for the flow processing system (superordinate concept) and its two subordinate concepts (BP (Business Process) processing system, job processing system).
The flow processing system is expressed by connecting “node calls” that are collective elements with “edges” that are connection terminals. One “node call” executes a process corresponding to one business unit of a flow by calling a “node” allocated on a computer different from the computer that executes the “node call”. Further, “partial flow” is obtained by sequentially extracting “node calls” of a route on one execution path from a start point to an end point.

  An example of the flow processing system is a BP processing system. In the BP processing system, “flow” is embodied as “BP”, “partial flow” as “partial BP”, “node call” as “service call activity”, and “node” as “service”. The

  Another example of the flow processing system is a job processing system. In the job processing system, “flow” is realized as “job net”, “partial flow” is realized as “partial job net”, “node call” is called “job call”, and “node” is realized as “job”. Is done.

  Heretofore, two examples of the flow processing system have been described. What is common in these flow processing systems is that when there are multiple computer candidates to which call destinations (nodes, services, jobs) are assigned, which call destination is assigned to which computer, high reliability can be obtained. This is a problem.

  For example, in a BP processing system, which is one of typical examples of a flow processing system, the reliability of a service called by executing an activity defined by a BP can be improved by an allocation method that takes the BP into consideration.

  For example, in the BP processing system, there are physical computers a1 and a2, and it is assumed that there is a service s1 executed by a virtual computer on the physical computer a1 and a service s2 executed by a virtual computer on the physical computer a2. If there is a predetermined BP that calls both services s1 and s2, if a failure occurs in either of the physical computers a1 and a2, the execution process of the predetermined BP cannot be continued.

  As described above, the reason why the reliability is lowered is that the number of physical computers that allocate resources to services called from the same BP is large. Therefore, reliability can be improved by allocating services called from the same BP to the same physical computer.

  However, in the BP processing system, when the total resource amount of the service called from a certain BP is larger than the resource amount of the allocation-destination physical computer, in the conventional technology, the total resource amount of the service called from the allocation target BP is larger. A new physical computer is prepared or a service is distributed to a plurality of physical computers without considering BP. In the former case, extra resources are generated in the physical computer, and in the latter case, reliability is lowered.

  Therefore, the present invention solves the above-described problem, and assigns each node called from a flow of a flow processing system (including a BP processing system and a job processing system) to a plurality of physical computers. The main purpose is to suppress the decrease in reliability.

In order to solve the above problems, the present invention provides a flow execution device for executing a flow, a node execution device for executing a node called from a node call of the flow, and a physical computer resource allocated to the executed node. A resource allocation method by a flow processing system configured to include an operation management device that manages resources of the node execution device,
The operation management device is
Reads from the storage unit correspondence information between a node call of the flow and a node called by execution of the node call, necessary resource amount information for each called node, and resource amount information available for each node execution device ,
For the flow, the necessary resource amount information of the node called by the execution of the flow is aggregated according to the correspondence information, the required resource amount of the flow is obtained, and the available resource amount exceeding the obtained required resource amount of the flow is calculated. Search for one or more node execution devices with
When one or more node execution devices are searched, one node execution device is selected from the searched one or more node execution devices, and the selected node execution device is selected from the flow. Assign the node to be called,
When no node execution device is searched, the flow is divided into a plurality of partial flows along the execution path for each branch of the flow, and a node called by executing the flow for each partial flow The necessary resource amount information is aggregated, the required resource amount of the partial flow is obtained, the node execution device having an available resource amount exceeding the obtained required resource amount of the partial flow is searched, and the searched node A node to be called for each partial flow is assigned to the execution device.
Other means will be described later.

  ADVANTAGE OF THE INVENTION According to this invention, when assigning each node called from the flow of a flow processing system to a some physical computer, the fall of the reliability of the execution path of a flow can be suppressed.

  Hereinafter, a first embodiment (BP processing system) and a second embodiment (job processing system) of a flow processing system to which the present invention is applied will be described. First, a first embodiment of a BP processing system will be described in detail with reference to the drawings.

FIG. 1 shows a block diagram of a BP processing system. The BP processing system is configured by connecting one or more BP execution devices 2, one or more operation management devices 1, and two or more service execution devices 3 via a network 8.
The BP execution device 2, which is a physical computer, includes a CPU 41, a network interface 42, a main storage device 45, and a secondary storage device 43. The network interface 42 is connected to the network 8.
Note that the operation management apparatus 1 and the service execution apparatus 3 are also configured as physical computers including the respective computer resources, like the BP execution apparatus 2. The main storage device 44 of the operation management device 1 is installed with a processing program (described later in FIG. 2) for the BP execution device 2 and the service execution device 3.

The operation management apparatus 1 instructs the BP execution apparatus 2 to execute BP, and instructs the service execution apparatus 3 to allocate a service and resources allocated to the service.
A resource is a computer resource of the service execution device 3, and is defined by, for example, the performance of a CPU, memory, or the like. The BP execution device 2 calls a service executed by the service execution device 3 by executing the BP.
The user who uses the BP requests the BP execution device 2 to execute the BP, and the BP execution device 2 transfers a request message to the service execution device 3 so as to execute the service corresponding to the BP through the network 8. The execution result of the service execution device 3 is transmitted to the BP execution device 2 through the network 8, and finally the execution result is transmitted to the user.

  Furthermore, the operation management apparatus 1 and the service execution apparatus 3 are connected by a management network 9 in order to control the power supply of the service execution apparatus 3 from the operation management apparatus 1. Since the network 8 and the management network 9 are different in the type of data that flows, each device connected to each network must have a separate interface for connecting to each network. Is desirable.

FIG. 2 is a configuration diagram showing functions of the BP processing system. The main storage device 44 of the operation management apparatus 1 includes an allocation planning unit 11, a BP instruction unit 12, a service instruction unit 13, and a power control instruction unit 14, and the OS 46 is activated.
The allocation planning unit 11 plans the resource allocation process to the service by editing the allocation copy table 15b based on the allocation master table 15a.
The BP instruction unit 12 instructs the BP management unit 21 to update the contents of the execution BP storage unit 22 based on the contents of the BP definition tables 16 a and 16 b via the network 8.
The service instruction unit 13 provides services on the virtual machine 32 based on the contents of the BP allocation tables 17a and 17b, the partial BP allocation tables 20a and 20b, the redundancy definition tables 19a and 19b, and the service allocation tables 18a and 18b. In addition, the virtual machine control unit 31 is instructed to allocate resources via the network 8.
The power control instruction unit 14 instructs the power control unit 35 to turn on and off the service execution device 3 via the management network 9.

The operation management apparatus 1 stores an allocation master table 15a and an allocation copy table 15b.
The allocation master table 15a includes a BP definition table 16a, a BP allocation table 17a, a service allocation table 18a, a redundancy definition table 19a, and a partial BP allocation table 20a.
The allocation copy table 15b includes a BP definition table 16b, a BP allocation table 17b, a service allocation table 18b, a redundancy definition table 19b, and a partial BP allocation table 20b.
The allocation copy table 15b is a copy created from the allocation master table 15a, and the copy is edited. When the editing result is normally reflected, the contents of the allocation copy table 15b are reflected in the allocation master table 15a.

The BP execution device 2 includes a BP management unit 21, an execution BP storage unit 22, a BP execution unit 23, and an OS 46.
The BP management unit 21 receives an instruction from the BP instruction unit 12 and updates the contents of the execution BP storage unit 22. The execution BP storage unit 22 stores a combination of activities that are parts for defining the flow of BP processing. In each activity, service invocation and conditional branching are performed.
The BP execution unit 23 calls a service according to the BP information in the execution BP storage unit 22.

  The service execution device 3 includes a virtual machine control unit 31, a virtual machine 32, a power supply control unit 35, and an OS 46. The virtual computer 32 includes a service execution unit 33 and a load monitoring unit 34.

The virtual computer control unit 31 receives an instruction command from the service instruction unit 13 and instructs the virtual computer 32 with the following instruction content.
Assign a service execution unit 33 to be executed.
• Allocate resources. The amount of resources to be allocated can be freely set within a range that does not exceed the resources of the service execution device 3.
The resource usage of the virtual machine 32 is acquired.

The virtual machine 32 allocates the necessary resource amount necessary for executing each service called by the execution of the BP activity executed by the BP execution unit 23 from the resources provided by the service execution device 3. Then, the service execution unit 33 is activated. Note that the resources provided by the virtual machine 32 from the service execution device 3 may be unchanged (fixed) with respect to time or may be variable.
The service execution unit 33 executes one or more services using resources allocated by the virtual machine 32.
The load monitoring unit 34 monitors the load given to the resources allocated to the service by the service execution unit 33 executing the service by periodically accessing the virtual machine control unit 31.
The power control unit 35 receives the power on / off instruction from the power control instruction unit 14 and controls the power of the service execution apparatus 3 that is the self apparatus.

  FIG. 3 is a configuration diagram showing the BP definition tables 16a and 16b of the BP processing system. The BP definition tables 16a and 16b store a BP name T1001, a BPIDT 1002, an activity IDT 1003, a service IDT 1004, a necessary resource amount T1005, and an allocation destination physical computer IDT 1006 in association with each other.

  The BP name T1001 uniquely identifies the BP.

  The BPIDT 1002 indicates a uniquely assigned ID indicating the entire BP and an ID for identifying each divided BP when the BP can be divided. Note that the ability to divide a BP means that a branch is defined in the BP definition, and that a BP can be divided as a partial BP for each execution path.

  When a branch is not defined in the BP, for example, only the BPID “BP1” indicating the entire BP is specified, such as T1007 whose BP name is “BP1”, but when a branch is defined in the BP, For example, as shown in T1008 to T1010 where the BP name is “BP3”, the BPID “BP3” indicating the entire BP, and the partial BP for each execution path are subsequently set to “BP3-1” T1009, “BP3-2” T1010. Identify as follows. In this example, the BPID indicating the entire BP and the BP name are the same, but different IDs may be allocated. In that case, it may be understood that the BPID indicates the entire BP.

  The activity IDT 1003 specifies an activity constituting the BP specified by the BP name T1001 and the BPIDT 1002. Therefore, for example, T1012 to T1014 whose activity ID is “A3” indicate the same activity.

  The service IDT 1004 specifies a service called from the activity of the activity IDT 1003. Therefore, T1011 having an activity ID “A1” and T1012 to T1014 having an activity ID “A3” indicate that the service “A” having the same service ID is called, although the activities are different.

The required resource amount T1005 is the resource amount of the service execution device 3 necessary for executing the service of the service IDT 1004. Therefore, regarding services called from the same activity on a plurality of execution paths, the amount of resources required to execute from each execution path is different from the amount of resources required by all BPs.
For example, the required resource amount corresponding to the activity ID “A3” is “20” in T1012 in the entire case, and the required resource amount required for executing the execution path of the partial BP “BP3-1” is “10” in T1013. "

That is, when activities (for example, “A3”) overlap between the partial BPs, the required resource amount (“20” of T1012) in the entire BP of the activity is the required resource amount (T1013 of T1013) in each partial BP of the activity. “10” and “10” of T1014).
In addition, as described above, the ratio of the required resource amount of the entire BP to the activity of each partial BP may be equally divided between the partial BPs, or by the execution of each partial BP. In proportion to the transaction amount to be processed, each partial BP may be different or a source amount may be allocated.

  The assignment destination physical computer IDT 1006 identifies the service execution device 3 (physical computer) to which the service ID service is assigned.

FIG. 4 is a configuration diagram showing the redundancy definition tables 19a and 19b of the BP processing system. The redundancy definition tables 19a and 19b manage the BPIDT 1002 that the system operator makes redundant. Redundant BPIDT 1101 (BP3-1) is defined before execution of the allocation plan. In addition, not only the partial BP (BP3-1) but also BP (BP4) can be specified as a target for redundancy.
The redundancy definition tables 19a and 19b are created by input from the user. As an input means from the user, the BPID may be input as a character string or selected from a list, or the BP and its activity are displayed in a graph (see FIG. 8) and designated by dragging the mouse from the graph display. You may input BP which exists in an area | region.

  FIG. 5 is a configuration diagram showing the BP allocation tables 17a and 17b of the BP processing system. The BP allocation tables 17a and 17b manage the sum of resources necessary for executing the service called from the BP indicated by the BPID or the partial BP.

The total resource amount T1201 is a total of resources necessary for executing the service called from the BP indicated by the BPIDT 1002. This total is obtained from the total required resource amount T1005 of records having the same BPID in the BP definition tables 16a and 16b.
For example, when BPID = “BP1”, “20 (A1)” + “20 (B1)” + “30 (C1)” = 70 is expressed as “total required resource amount” when expressed as “required resource amount (activity ID)”. (T1205). Similarly, in the case of the partial BP, for example, when BPID = “BP3-1”, “10 (A3)” + “20 (D3)” = “30” is the total resource amount.

  Allocated flag T1202 indicates whether a resource is allocated to the BP indicated by BPIDT 1002 (true indicating true) or not (false indicating false). For example, when “false” is indicated as in T1206, this indicates a state where resources are not yet allocated.

  The distributed allocation flag T1203 indicates whether the resource allocated to the BP indicated by the BPIDT 1002 is distributed and allocated to a plurality of physical computers (true: distributed allocation), and is allocated concentrated on one physical computer. (False: centralized allocation). When “false” is indicated as in T1207, it indicates that centralized allocation is performed.

  The assignable physical computer list T1204 shows a list of physical computers to which resources can be assigned. T1208 indicates that there are three “X”, “Y”, and “Z” as a list of physical computers.

  FIG. 6 is a configuration diagram showing the partial BP allocation tables 20a and 20b of the BP processing system. The partial BP allocation tables 20a and 20b extract only the information on the partial BP when the information on the partial BP is defined in the BP allocation tables 17a and 17b shown in FIG. The BP name T1001 is a BP name to which the partial BP belongs, and the BPIDT 1002 specifies the BPID in the partial BP of each BP.

The total resource amount T1301 is the total resource amount of the BPID of the corresponding partial BP. The calculation method is the same as the total resource amount T1201 in FIG.
Allocated flag T1302 indicates whether resources are allocated to the partial BP indicated by BPIDT 1002 (true indicating true) or not (false indicating false).
The assignable physical computer list T1303 shows a list of physical computers to which resources can be assigned.
FIG. 7 is a configuration diagram showing service allocation tables 18a and 18b of the BP processing system. The service allocation tables 18a and 18b store the physical computer IDT 1401, the resource amount T1402, the free resource amount T1403, and the allocation list T1404 in association with each other.

  The physical computer IDT 1401 identifies the service execution device 3.

  The resource amount T1402 indicates the total amount of resources that can be allocated in the physical computer with the physical computer IDT1401.

  The free resource amount T1403 indicates the remaining resource amount obtained by removing the already allocated resource amount from the resource amount in the physical computer of the physical computer IDT1401. For example, if there is a resource amount of “100” in a predetermined physical computer, and the resource amount of “40” has already been allocated, the free resource amount is “60”.

  The allocation list T1404 is a list of services already allocated to the physical computer with the physical computer ID. For example, the notation “BP1” of T1405 indicates that the BP service indicated by the ID is assigned to “X” indicated by the physical computer ID. Since the resource amount used by BP1 is “70” (described as “BP1 (70)” in the figure), the available resource amount is the resource amount “100” that can be allocated—the resource of “BP1” that is actually allocated. The quantity “70” = “30”.

  FIG. 8 is an explanatory diagram showing two types of service allocation methods (centralized allocation and distributed allocation). In this explanatory diagram, solid arrows indicate the order in which activities are executed according to the execution path, and dotted arrows indicate call relationships from activities to services.

The BP execution unit 23 executes BP “1” and BP “4”. The BP “1” is executed in the order of activities “A1” → “B1” → “C1”. In the course of this execution, for example, the activity “A1” calls the service “A” executed by the service execution unit 33 of the service execution apparatus 3 (dotted line arrow).
Note that a conditional branch is defined in BP “4” on the right side, the partial BP “4-2” indicating the execution path of the activity “D4” → “F4”, “D4” → “G4” → “ And a partial BP “4-1” indicating an execution path of “H4”. Then, the partial BP “4-1” or the partial BP “4-2” are executed in the order of the activity of any execution path.

  First, the centralized allocation 51 indicates that services called by executing all activities belonging to a predetermined BP are concentrated and allocated to the same service execution device 3. For example, the activity “A1, B1, C1” of BP “1” is a centralized assignment because the service “A, B, C” assigned to the same service execution device 3 is called.

On the other hand, the distributed allocation 52 indicates that a service called by executing all activities belonging to a predetermined BP is distributed and allocated to a plurality of service execution apparatuses 3. When distributed and allocated, a service group called by executing the partial BP according to the execution path is set as a minimum unit to be allocated.
For example, in the activity “D4, F4, G4, H4” of the BP “4”, the partial BP “4-1” is assigned to the first service execution apparatus 3, and the partial BP “4-2” is 2 Since it is assigned to the third service execution apparatus 3, it is distributed assignment.

  As described above, the execution path described with reference to FIG. 8 may be input by the user in advance, or may be mechanically extracted using the entire BP as an input. For example, a BPEL (Business Process Execution Language) BP is an XML (Extensible Markup Language) tree structure having an activity as a node. A method of extracting an execution path by tracing in order from (point) can be mentioned.

  FIG. 9 is an explanatory diagram showing two types of service allocation methods (centralized allocation and redundant allocation). Compared to FIG. 8, the difference is that the activity “F4” uses the service “D” on the two service execution apparatuses 3.

In the BP processing system, it is conceivable to consider service redundancy in order to increase reliability. In that case, even if only one service is made redundant, it is not sufficient in the BP processing system, and at least all the services called by executing the activity on the execution path on which the service is executed must be made redundant. Therefore, the minimum unit of redundant allocation is a partial BP (BP “4-2” in the figure).
On the other hand, the service of the entire BP may be made redundant. In this case, a failure to all partial BPs of the entire BP can be dealt with instead of requiring many resources.

Redundant allocation means that under the situation where the entire BP has already been allocated, at least a part of the allocated BP is transferred to a service execution apparatus 3 other than the allocated service execution apparatus 3. This is a redundant allocation method.
For example, BP “4” has already been assigned to the right service execution apparatus 3. Further, in order to increase the reliability of the BP “4”, the redundant allocation 53 of the BP “4-2” is also made to the service execution device 3 on the left side.

  FIG. 10 is a flowchart showing the concentrated resource allocation processing executed by the allocation planning unit 11.

  First, the allocation master table 15a is copied to the allocation copy table 15b (S101). Specifically, from the BP definition table 16a to the BP definition table 16b, from the BP assignment table 17a to the BP assignment table 17b, from the service assignment table 18a to the service assignment table 18b, and from the redundancy definition table 19a to the redundancy definition table 19b. The partial BP allocation table 20a is copied to the partial BP allocation table 20b.

  Next, referring to the BP allocation table 17b, a loop (S111 to S119) for selecting the allocation target BP from the BP to which the resource is not allocated or the partial BP in the descending order of the total resource amount is executed. By executing this loop, the allocated flag of the allocation target BP is updated from “false” to “true”. By assigning BPs in descending order of the total amount of resources, it is possible to prevent resource vacancies from being fragmented into a plurality of physical computers.

With reference to the service allocation table 18b, a physical computer having a free resource amount exceeding the total resource amount of the allocation target BP is searched (S112).
First, as a result of S112, if a physical computer capable of centralized allocation is not found (S113, No), the process proceeds to S116.
If the allocation target BP is not a partial BP (S116, Yes), the allocation target BP is distributed and allocated to a plurality of physical computers (S117). Specifically, the process of FIG. 10 described later is called as a subroutine.
If the allocation target BP is a partial BP (No in S116), an error notification is sent that there is a shortage of resources because no further distributed allocation is possible (S118).

  On the other hand, when a physical computer is searched as a result of S112 (S113, Yes), when the same BP is not allocated to the physical computer (S114, Yes), the allocation target BP is centrally allocated to the physical computer. (S115), the allocation target BP loop is terminated (S119).

  If the same BP is assigned to the physical computer (No at S114), the physical computer is deleted from the assignable physical computer list of the BP, and a physical computer that can be accommodated again is searched (S112). It is assumed that resources are allocated to a physical computer having the largest amount of free resources among the physical computers searched in general.

  For the centralized allocation (S115), specifically, the allocation target BP is allocated to the allocation target physical computer specified in S112. That is, the ID of the identified physical computer to be allocated is written into the allocation destination physical computer ID column in the record indicating the allocation target BP in the BP definition table 16b. Then, the BPID of the allocation target BP is written in the allocation list column in the record indicating the specified allocation target physical computer in the service allocation table 18b. “False” is set in the distributed allocation flag corresponding to the allocation target BP in the BP allocation table 17b.

  Further, after completing the loop in S119, resources are allocated to the service according to the contents of the allocation copy table 15b updated in S114, S115, and S117 (details will be described later in FIG. 11) (S121). This process may be executed after the user confirms the contents of the allocation copy table 15b and obtains permission from the user. In order to realize the process of allocating resources to the service, the service instruction unit 13 causes the virtual machine 32 to allocate resources used by the virtual machine 32 via the virtual machine control unit 31, and activates the service execution unit 33. The assigned service is executed using the assigned resource.

  Based on the allocation result, the latest allocation state is reflected from the allocation copy table 15b to the allocation master table 15a (S122). For example, when all the contents of the allocation copy table 15b are normally allocated, the contents of the allocation master table 15a are copies of the allocation copy table 15b. On the other hand, when an allocation error (S118, S209 in FIG. 11 described later) occurs due to a resource shortage or the like in part of the contents of the allocation copy table 15b, the contents of the allocation master table 15a are allocated due to the allocation error. The writing of the contents of the allocation copy table 15b that has not occurred is omitted. Since the allocation copy table 15b is unnecessary, it is deleted.

  The unused physical computer is turned off (S123). Specifically, the allocation list of the service allocation table 18a is referred to, and power saving is executed by turning off the power of a physical computer to which resources are not allocated. Therefore, the power control instruction unit 14 instructs the power control unit 35 of the physical computer to which no resource is assigned via the management network 9 to turn off the physical computer. The process of FIG. 10 has been described above.

  FIG. 11 is a flowchart for distributed allocation (S117) of the allocation target BP shown in FIG.

  First, “true” is set to the distributed allocation flag corresponding to the allocation target BP (S201).

  Next, a loop for referring to the partial BP allocation table 20b and selecting the allocation target BP in descending order of the total resource amount of the partial BP from the partial BP to which the resource corresponding to the allocation target BP is not allocated (S202 to S208). Execute. By executing this loop, the allocation flag of the allocation target BP in the partial BP allocation table 20b is updated from “false” to “true”. If the assigned flag of all the partial BPs corresponding to the assignment target BP becomes “true”, the assigned flag in the BP assignment table 17 is updated to “true”.

With reference to the service allocation table 18b, a physical computer having a free resource amount exceeding the total resource amount of the allocation target partial BP (in other words, a partial BP can be allocated) is searched (S203). In S204, the search result of S203 is determined.
First, if the physical computer is not searched (S204, No), the partial BP allocation work is terminated, assuming that the partial BP could not be allocated, and the process proceeds to S207.

On the other hand, if a physical computer is searched (S204, Yes), it is confirmed whether a BP including the allocation target part BP or the same partial BP is assigned to the physical computer, and if it is not assigned (S205, Yes), the allocation target part BP is allocated to the physical computer (S206).
On the contrary, it is confirmed whether or not the BP including the allocation target part BP or the same partial BP is allocated to the physical computer. If the BP is allocated (No in S205), the allocable physical computer list of the BP is confirmed. Then, the physical computer is deleted, and a physical computer that can accommodate the partial BP is searched again (S203). It is assumed that resources are allocated to a physical computer having the largest amount of free resources among the physical computers searched in general.

  Here, it is determined whether or not the allocation of all the partial BPs belonging to the allocation target BP is completed (S207). The completion of allocation of all partial BPs belonging to the allocation target BP indicates that all the services called by execution of the allocation target BP have been allocated by the allocation processing in S206. In the case of Yes in S207, the assigned flag in the BP assignment table 17b is updated to “true”, and the process returns from the subroutine to the caller (S117).

  On the other hand, if the allocation of the allocation target BP is not completed (S207, No), the loop is terminated (S208).

  If the loop of S202 to S208 is exited, the answer to S207 is not Yes, so the partial BP resource is not allocated due to insufficient resources, and an error that the service cannot be executed is notified to the user and the caller (S209). ). After these processes are completed, the process returns from the subroutine to the caller (S117).

  FIG. 12 is a block diagram showing the service allocation table 18 describing the BP distributed allocation result. It is a result of the service allocation table when BP1, BP2, and BP3 in the BP definition table 16b are allocated to two physical computers X and Y. BP3 cannot be centrally allocated, and is distributedly allocated to BP3-1 and BP3-2. In the conventional technology, the user must select whether to prepare the third physical computer Z or to deploy without regard to the execution path. The former causes waste of physical computer resources, and the latter Although the reliability decreased, resource allocation with improved reliability could be realized without waste by the distributed allocation processing (see FIG. 11) of the present embodiment.

  FIG. 13 is a block diagram showing the service allocation table 18 describing the BP distributed allocation arrangement result. This is a result of the service allocation table when BP4 and BP5 in the BP definition table 16b are allocated to three physical computers X, Y, and Z. Neither BP4 nor BP5 can be centrally allocated, BP4 is distributed and allocated to BP4-1 and BP4-2, and BP5 is distributed and allocated to BP5-1 and BP5-2. In the case where the total resource amount of a certain BP is larger than the resource amount of one physical computer, this embodiment can allocate resources without impairing reliability.

FIG. 14 is a block diagram showing the service allocation table 18 describing the redundant allocation allocation result of BP. This is a result of the service allocation table when BP1, BP2, and BP3 in the BP definition table 16b are allocated to three physical computers X, Y, and Z. However, the user designates that the partial BP of BP3-1 is made redundant in the redundancy definition table 19b.
BP3-1 is not assigned to the physical computer Z to which BP3 including the same partial BP is assigned, but is assigned to another physical computer. As a result, even if the physical computer Z fails, for example, if the physical computer Y is operating, the execution path of the BP3-1 can be saved. As a result of this embodiment, the BP was successfully made redundant.

FIG. 15 is a configuration diagram showing the service allocation table 18 describing the BP distributed allocation arrangement and the redundant allocation arrangement result. This is a result of the service allocation table when BP1, BP3, and BP5 in the BP definition table 16b are allocated to three physical computers X, Y, and Z. However, the user designates that the partial BP of BP3-1 is made redundant in the redundancy definition table 19b.
Thus, depending on the result of the resource amount of the physical computer and the total resource amount of BP, distributed allocation can be realized simultaneously with redundant allocation of BP.
In the BP processing system, there are cases where the total amount of resources of services called by the execution of the BP activity is large, or redundancy of services is absolutely necessary. In such a BP processing system, an allocation system with improved reliability could be realized.

  According to this embodiment, the reliability of the service called by execution of BP improves. For example, when a certain BP is divided into two parts BPb1 and b2, and a service called from b1 and a service called from b2 are distributed to two service execution devices a1 and a2, the service execution device of a1 fails. However, the remaining part BPb2 can be executed. Further, since the BP can be finely divided for each execution path, waste of resources can be omitted.

The second embodiment (job processing system) of the flow processing system to which the present invention is applied will be described below. As already described with reference to FIG. 20, the difference between the first embodiment (BP processing system) and the second embodiment (job processing system) is that the data content to be handled is BP or a job. Yes, each technical feature (centralized allocation, distributed allocation, redundant allocation) of the first embodiment is also included in the second embodiment.
Therefore, in order to clarify the correspondence between the constituent elements of the first embodiment and the second embodiment, the last two digits “12” of the constituent elements of the first embodiment (for example, the BP instruction unit 12) and the first The code is configured such that the last two digits “12” of the component of the second embodiment (for example, the job net instruction unit 112) match.

  FIG. 16 shows a configuration diagram of the job processing system, similar to FIG. The operation management apparatus 101 issues a job net execution instruction to the job net management apparatus 102 and allocates resources and jobs to the job execution apparatus 103.

FIG. 17 is a block diagram showing functions of the job processing system, as in FIG. The main storage device 44 of the operation management apparatus 101 includes an allocation planning unit 11, a job net instruction unit 112, a job instruction unit 113, and a power control instruction unit 14, and the OS 46 is activated.
The allocation planning unit 11 plans the resource allocation process to the job by editing the allocation copy table 115b based on the allocation master table 115a.
The job net instruction unit 112 instructs the job net management unit 121 to update the contents of the execution job net storage unit 73 based on the contents of the job net definition tables 116 a and 116 b via the network 8.
The job instruction unit 113 runs on the virtual computer 32 based on the contents of the job net allocation tables 117a and 117b, the partial job net allocation tables 120a and 120b, the redundancy definition tables 119a and 119b, and the job allocation tables 118a and 118b. Is assigned to the virtual machine control unit 31 via the network 8.
The power control instruction unit 14 instructs the power control unit 35 to turn on / off the job execution apparatus 103 via the management network 9.

The operation management apparatus 101 stores an allocation master table 115a and an allocation copy table 115b.
The allocation master table 115a includes a job net definition table 116a, a job net allocation table 117a, a job allocation table 118a, a redundancy definition table 119a, and a partial job net allocation table 120a.
The allocation copy table 115b includes a job net definition table 116b, a job net allocation table 117b, a job allocation table 118b, a redundancy definition table 119b, and a partial job net allocation table 120b.
The allocation copy table 115b is a copy created from the allocation master table 115a, and the copy is edited. When the editing result is normally reflected, the contents of the allocation copy table 115b are reflected in the allocation master table 115a.

  FIG. 18 is an explanatory diagram showing a job net job assignment method, similar to FIG. The job net execution unit 123 executes the job net “1” and the job net “4”. The centralized allocation and distributed allocation of the job processing system in FIG. 18 are the same ideas as the BP processing system. When a job is arranged in one job execution apparatus 103, centralized assignment is performed. When a job is arranged in a plurality of job execution apparatuses 103, distributed assignment is performed.

FIG. 19 is an explanatory diagram showing a job net assignment method, similar to FIG. The job net “1” is centrally allocated to the same one job execution apparatus 103, and the job net “4” is centrally allocated to another job execution apparatus 103.
Here, consider making a specific job redundant. Considering execution of job processing, if jobs are not independent processes but are dependent on each other (for example, the result of job call “D4” affects job call “F4”), one job Even if only redundancy is made, a problem arises in reliability. Therefore, all actual job programs called from jobs on the execution path including a specific job on the job net to be made redundant must be made redundant.
Redundant allocation 53 is made redundant in units of partial job net “4-2” indicating the execution path of job “D4, F4” of job net “4” and allocated to one job execution apparatus 103.

  As described above, for each of the first embodiment (BP processing system) and the second embodiment (job processing system), the reliability of the flow processing system is improved and the waste of resources of the physical computer is minimized. In addition, a service or job called from the management apparatus can be assigned to the resource of the physical computer.

FIG. 21 is a table summarizing the allocation method in the first embodiment (BP processing system). Hereinafter, an outline of processing of the operation management apparatus 1 will be described.
The required resource amount increases in the order of BP (whole)> partial BP> (one) service call activity.
When allocating the BP to be allocated, if the entire BP can be centrally allocated to one unit, high reliability can be obtained.
If the entire BP is too large to be centrally allocated, distributed allocation is attempted for a plurality of units (a small number) with a partial BP composed of one or more activities as a unit. When distributed allocation of partial BPs is possible, it is allocated to one unit in the execution path unit, and appropriate reliability can be obtained.
Furthermore, even if the partial BP is used as a unit, if distributed allocation cannot be performed, an error is issued instead of assigning one more detailed activity as a unit.

  Furthermore, when the centralized allocation of the entire BP or the distributed allocation of the partial BPs can be performed, redundant allocation may be performed in order to further improve the reliability of the allocated BP. In the redundant allocation, the entire BP designated by the user in advance or the service execution device 3 different from the allocated service execution device 3 is newly searched in units of partial BPs, or the entire designated BP, or , Assign a partial BP.

It is a block diagram which shows the BP processing system regarding 1st Embodiment of this invention. It is a block diagram which shows the function of the BP processing system regarding 1st Embodiment of this invention. It is a block diagram which shows the BP definition table of the BP processing system regarding 1st Embodiment of this invention. It is a block diagram which shows the redundancy definition table of the BP processing system regarding 1st Embodiment of this invention. It is a block diagram which shows the BP allocation table of the BP processing system regarding 1st Embodiment of this invention. It is a block diagram which shows the partial BP allocation table of the BP processing system regarding 1st Embodiment of this invention. It is a block diagram which shows the service allocation table of the BP processing system regarding 1st Embodiment of this invention. It is explanatory drawing which shows the allocation method (centralized allocation, distributed allocation) of two types of services in the BP processing system regarding 1st Embodiment of this invention. It is explanatory drawing which shows the allocation method (centralized allocation and redundant allocation) of two types of services in the BP processing system regarding 1st Embodiment of this invention. It is a flowchart which shows the resource allocation process regarding 1st Embodiment of this invention. It is a flowchart which shows the process which carries out the distributed allocation regarding 1st Embodiment of this invention. It is a block diagram which shows the service allocation table which describes the distributed allocation arrangement | positioning result of BP regarding 1st Embodiment of this invention. It is a block diagram which shows the service allocation table which describes the distributed allocation arrangement | positioning result of BP regarding 1st Embodiment of this invention. It is a block diagram which shows the service allocation table which describes the redundant allocation arrangement | positioning result of BP regarding 1st Embodiment of this invention. It is a block diagram which shows the service allocation table which describes the distributed allocation arrangement | positioning and redundant allocation arrangement | positioning of BP regarding 1st Embodiment of this invention. It is a block diagram which shows the job processing system regarding 2nd Embodiment of this invention. It is a block diagram which shows the function of the job processing system regarding 2nd Embodiment of this invention. It is explanatory drawing which shows the allocation method (centralized allocation, distributed allocation) of two types of services in the job processing system regarding 2nd Embodiment of this invention. It is explanatory drawing which shows the allocation method (centralized allocation and redundant allocation) of two types of services in the job processing system regarding 2nd Embodiment of this invention. It is explanatory drawing which shows the term of a flow processing system. It is explanatory drawing which shows three types of allocation systems regarding one Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Operation management apparatus 2 BP execution apparatus 3 Service execution apparatus 8 Network 9 Management network 11 Assignment plan part 12 BP instruction part 13 Service instruction part 14 Power supply control instruction part 15a Assignment master table 15b Assignment copy table 16 BP definition table 17 BP allocation table 18 Service allocation table 19 Redundancy definition table 20 Partial BP allocation table 21 BP management unit 22 Execution BP storage unit 23 BP execution unit 31 Virtual computer control unit 32 Virtual computer 33 Service execution unit 34 Load monitoring unit 35 Power control Part 41 CPU
42 Network interface 43 Secondary storage device 44 Main storage device 45 Main storage device 46 OS
51 Centralized allocation 52 Distributed allocation 53 Partial redundant allocation 101 Operation management apparatus 102 Job net management apparatus 103 Job execution apparatus 112 Job net instruction section 113 Job instruction section 115a Allocation master table 115b Allocation copy table 116 Job net definition table 117 Job net Allocation table 118 Job allocation table 119 Redundancy definition table 120 Partial job net allocation table 121 Job net management unit 122 Job net storage unit for execution 123 Job net execution unit 133 Job execution unit

Claims (9)

A flow execution device that executes a flow, a node execution device that executes a node that is called from a node call of the flow, and an operation that manages resources of the node execution device that are physical computer resources allocated to the executed node A resource allocation method by a flow processing system configured to include a management device,
The operation management device is
Reads from the storage unit correspondence information between a node call of the flow and a node called by execution of the node call, necessary resource amount information for each called node, and resource amount information available for each node execution device ,
For the flow, the necessary resource amount information of the node called by the execution of the flow is aggregated according to the correspondence information, the required resource amount of the flow is obtained, and the available resource amount exceeding the obtained required resource amount of the flow is calculated. Search for one or more node execution devices with
When one or more node execution devices are searched, one node execution device is selected from the searched one or more node execution devices, and the selected node execution device is selected from the flow. Assign the node to be called,
When no node execution device is searched, the flow is divided into a plurality of partial flows along the execution path for each branch of the flow, and a node called by executing the flow for each partial flow The necessary resource amount information is aggregated, the required resource amount of the partial flow is obtained, the node execution device having an available resource amount exceeding the obtained required resource amount of the partial flow is searched, and the searched node A resource allocation method, wherein a node to be called for each partial flow is allocated to an execution device. The flow processing system is configured as a BP processing system in which a BP (Business Process) execution device executes a BP as a flow and a service execution device executes a service as a node,
The partial flow is configured as a partial BP configured along an execution path for each branch of the BP,
The resource allocation method according to claim 1, wherein the node call is configured as a BP service call activity. The flow processing system is configured as a job net processing system in which a job net execution device executes a job net as a flow, and the job execution device executes a job as a node.
The partial flow is configured as a partial job net configured along an execution path for each branch of the job net,
The resource allocation method according to claim 1, wherein the node call is configured as a job call of a job net. In the step of assigning a node called from the flow to one node execution device, selecting the node execution device having the largest amount of resources that can be provided among the one or more node execution devices searched. The resource allocation method according to any one of claims 1 to 3. The operation management apparatus assigns a flow to a predetermined node execution apparatus, and when the flow or the partial flow is designated as redundant, the redundant designated flow or the partial flow is The resource allocation method according to any one of claims 1 to 4, wherein allocation is performed to the node execution device different from the predetermined node execution device. The operation management device assigns each partial flow composed of a predetermined flow to the predetermined node execution device group, and if the flow or the partial flow is designated as redundant, the redundancy is designated. The resource allocation according to any one of claims 1 to 4, wherein a flow or a partial flow thereof is allocated to the node execution device different from the predetermined node execution device group. Method. 2. In the step when the node execution device is not searched, when a node cannot be allocated to the node execution device in units of partial flows, an error notification is given for resource shortage. The resource allocation method according to claim 6.   A resource allocation program for causing each device of the flow processing system to execute the resource allocation method according to any one of claims 1 to 7. A flow execution device that executes a flow, a node execution device that executes a node that is called from a node call of the flow, and an operation that manages resources of the node execution device that are physical computer resources allocated to the executed node A flow processing system including a management device,
The operation management device is
Reads from the storage unit correspondence information between a node call of the flow and a node called by execution of the node call, necessary resource amount information for each called node, and resource amount information available for each node execution device ,
For the flow, the necessary resource amount information of the node called by the execution of the flow is aggregated according to the correspondence information, the required resource amount of the flow is obtained, and the available resource amount exceeding the obtained required resource amount of the flow is calculated. Search for one or more node execution devices with
When one or more node execution devices are searched, one node execution device is selected from the searched one or more node execution devices, and the selected node execution device is selected from the flow. Assign the node to be called,
When none of the node execution devices are searched, the flow is divided into a plurality of partial flows along the execution path for each branch of the flow, and a node called by executing the flow for each partial flow The necessary resource amount information is aggregated, the required resource amount of the partial flow is obtained, the node execution device having an available resource amount exceeding the obtained required resource amount of the partial flow is searched, and the searched node A flow processing system, wherein a node to be called for each partial flow is assigned to an execution device.

Images