JP2016091508A - Load distribution device, load distribution program, and processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a load distribution device capable of distributing a plurality of resources so that the resources are efficiently executed.SOLUTION: A load distribution device according to the present invention comprises: a storage part stored with information representing the maximum number of requests which can be processed by a plurality of request processing execution parts all together for each request kind, information representing the maximum number of requests which can be processed by an individual request processing execution part for each request kind, information associated with request processing execution parts to which request processing based upon a specific request kind is preferentially distributed, and information representing the number of requests being processed by a request processing execution part for each request kind; and a distribution destination determination part which determines a distribution destination among a plurality of request processing execution parts on the basis of a request kind included in a received request signal and those pieces of information stored in the storage part.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a load distribution device, a load distribution program, and a processing device, and can be applied to, for example, an SBC (Session Order Controller) device.

  2. Description of the Related Art Conventionally, with the development of network and computer technology, information processing apparatuses tend to adopt a configuration in which a plurality of resources are connected in parallel in order to process a large number of requests, thereby increasing the scale. Such a configuration also causes a problem that a load is concentrated on a specific resource when a certain request is processed. In order to solve such a problem, various techniques for distributing the load have been proposed (see Patent Document 1).

  For example, a load balancer (load balancer) distributes resources to be processed by the following eight methods and performs load balancing control of a plurality of resources. The methods (1) to (3) are static load distribution methods, and the methods (4) to (8) are dynamic load distribution methods. (1) The round robin method is a method for sequentially allocating resources. (2) The priority method is a method in which priorities are defined for resources and resources are allocated according to the priorities. (3) The ratio method is a method in which weighting is defined for resources and resources are allocated according to the ratio. (4) The least connection method is a method of monitoring the current allocation status for each resource and allocating it to the resource having the smallest allocation number. (5) The first test method is a method in which the current resource allocation status is monitored and allocated to resources determined to be processable at the fastest speed. (6) The observed method is a method for allocating resources to a resource having the smallest number of allocations within the corresponding period by monitoring the allocation status for a certain period. (7) The predictive method is a method for preferentially allocating resources that tend to decrease based on the result of monitoring the allocation status for a certain period and predicting the trend from the result. (8) The dynamic ratio method is a method of allocating resources having the lowest usage rate such as a CPU and a memory by monitoring the execution status of each resource.

JP 2010-122773 A

  However, when the load distribution target resource is, for example, a DSP (digital signal processor) capable of processing a plurality of functions, the load distribution apparatus assigns a request to each resource to execute the same function as much as possible. High execution efficiency can be obtained. Further, as the load distribution method when the processing load per resource and the allowable number of allocation are different for each requested function, the general load distribution methods (1) to (8) described above are not applicable. There was a point.

  Specifically, since the general load distribution method described above is based on the premise that any resource may be allocated if only the load amount is distributed, the same resource is preferentially selected for the same function. I can't. To solve this problem, when the load balancer always selects a fixed resource according to the request type, if the number of resources is less than the number of request types, only a specific request type is preferentially assigned. There is also a possibility that other subsequent request types cannot secure resources. In that case, the load balancer must also consider that resources are distributed such that a plurality of functions are executed to some extent so that a plurality of request types are processed with a single resource.

  Therefore, a load distribution device, a load distribution program, and a processing device that can distribute each resource so as to efficiently execute a plurality of resources are desired.

  A first aspect of the present invention is a load distribution apparatus that is connected to a plurality of request processing execution units that perform processing based on a request type included in a received request signal, and distributes the processing to any one of the request processing execution units. (1) information indicating the maximum number of requests that can be processed by the plurality of request processing execution units as a whole for each request type, information indicating the maximum number of requests that can be processed individually by the request processing execution unit for each request type, A storage unit for storing information on the request processing execution unit that preferentially distributes request processing based on a specific request type, and information indicating the number of requests being processed by the request processing execution unit for each request type; ) A distribution destination determination unit that determines a distribution destination to one of the plurality of request processing execution units based on the request type included in the received request signal and the information stored in the storage unit. It is characterized by .

  A load distribution program according to a second aspect of the present invention is connected to a plurality of request processing execution units that perform processing based on a request type included in a received request signal, and distributes the processing to one of the request processing execution units The computer mounted on the apparatus requests (1) information indicating the maximum number of requests that can be processed by the plurality of request processing execution units for each request type and the maximum number of requests that can be processed individually by the request processing execution units. Stores information shown for each type, information about the request processing execution unit that preferentially distributes request processing based on a specific request type, and information indicating the number of requests being processed by the request processing execution unit for each request type And (2) determining a distribution destination to one of the plurality of request processing execution units based on the request type included in the received request signal and the information stored in the storage unit. Wherein the function as the distribution destination determining unit.

  A third aspect of the present invention is a processing apparatus including the load distribution apparatus according to the first aspect of the present invention, wherein a request signal receiving unit that determines whether or not a received request signal can be received, and a request type included in the request signal And a plurality of request processing execution units that perform processing based on the above.

  According to the present invention, each resource can be distributed so that a plurality of resources can be efficiently executed.

It is a block diagram which shows the functional structure of the request signal processing apparatus which concerns on 1st Embodiment. It is explanatory drawing which shows the specification definition table | surface for every request classification which concerns on 1st Embodiment. It is explanatory drawing which shows the whole apparatus execution state table which concerns on 1st Embodiment. It is explanatory drawing which shows the resource definition table which concerns on 1st Embodiment. It is explanatory drawing which shows the execution state table | surface for every resource which concerns on 1st Embodiment. It is a flowchart which shows operation | movement (request reception) of the request signal processing apparatus which concerns on 1st Embodiment. It is a flowchart which shows operation | movement (load distribution) of the request signal processing apparatus which concerns on 1st Embodiment. It is a flowchart which shows operation | movement (1st list determination process) of the request signal processing apparatus which concerns on 1st Embodiment. It is a flowchart which shows operation | movement (2nd list determination process) of the request signal processing apparatus which concerns on 1st Embodiment.

(A) First Embodiment Hereinafter, a first embodiment of a load distribution apparatus, a load distribution program, and a processing apparatus according to the present invention will be described in detail with reference to the drawings.

(A-1) Configuration of First Embodiment FIG. 1 is a block diagram showing a functional configuration of a request signal processing apparatus 101 according to the first embodiment.

  In FIG. 1, the request signal processing apparatus 101 includes a request signal receiving unit 102, a load distribution function unit 103, a request processing execution unit 104 (104-0 to 104 -N), and a storage unit 105.

  The request signal processing apparatus 101 receives a request signal 106 from the outside and executes a corresponding process.

  The storage unit 105 is a storage unit that stores various data used in the processing of the request signal reception unit 102 or the load distribution function unit 103. In FIG. 1, as data stored in the storage unit 105, a request type specification definition table 201, an entire apparatus execution status table 301, a resource definition table 401, and a resource execution status table 501 are illustrated. .

  FIG. 2 is an explanatory diagram illustrating a specification definition table 201 (201-0 to 201-M) for each request type. In FIG. 2, a specification definition table 201 for each request type is allowed for each request type (# 0 to #M) and the number of requests that can be accepted by the entire apparatus (MEPRS: Maximum Executable Requests Number Per System) and individual resources. The number of possible requests (MERPR: Maximum Executable Requests number Per Resource) is included as a component.

  The request type means a request function type that can identify information included in the received request signal 106. In this embodiment, the specific contents are not limited. For example, for the purpose of distributing media signal processing of IP calls using different media codec signals for each session, the request type is an SIP signal. This corresponds to media codec type information extracted as a result of the negotiation.

  In FIG. 2, the specification definition table 201 for each request type defines a maximum request allowable number (entire apparatus) MEMRS and a maximum request allowable number (for each resource) MERPR for each request type.

  FIG. 3 is an explanatory diagram showing the entire apparatus execution state table 301.

  In FIG. 3, the entire apparatus execution status table 301 indicates the number of requests being executed for each request type being executed in the entire apparatus according to the request type (# 0 to #M).

  FIG. 4 is an explanatory diagram showing the resource definition table 401.

  In FIG. 4, a resource definition table 401 (401-0 to 401-N) includes an allocation method for defining whether only a specific request type is allocated to each resource or an arbitrary request type can be allocated, and an allocation method Includes a fixed allocation request type for specifying a request type to be allocated to the corresponding resource as a constituent element. The resource definition table 401 exists according to the number of resources.

  FIG. 5 is an explanatory diagram of an execution status table 501 for each resource.

  In FIG. 5, a per-resource execution state table 501 (501-0 to 501-N) indicates the number of requests for each corresponding request type being executed for each resource. In FIG. 5, the per-resource execution state table 501 indicates the number of requests being executed on the resource for each request type (# 0 to #M). This per-resource execution state table 501 exists according to the number of resources.

  In this embodiment, the resource is the request processing execution unit 104.

  The request signal receiving unit 102 receives the request signal 106 and determines whether the request can be received. Specifically, the request signal accepting unit 102 sets various specifications for each corresponding request type from among a plurality of predefined request type specification definition tables 201 based on the request types included in the received request signal 106. The original definition table 201 is determined, and the maximum allowable request number (entire apparatus) MEPRS is extracted from the specification definition table 201 for each corresponding request type. Then, the request signal receiving unit 102 compares the number of requests being executed for each request type stored in the overall apparatus execution table 301 to determine whether or not the request can be received. Detailed operation will be clarified in the operation section described later.

  The load distribution function unit 103 performs processing to distribute and distribute the request signal 106 determined to be acceptable by the request signal accepting unit 102 to the plurality of request processing execution units 104. Specifically, in the load distribution function unit 103, based on the request type included in the received request signal 106, the maximum allowable request number (per resource) MERPR, the resource definition table 401, and the per-resource execution state table 501 And select a resource to which the request is to be allocated. Detailed operation will be clarified in the operation section described later.

(A-2) Operation of the First Embodiment Next, the operation of the request signal processing apparatus 101 of the first embodiment will be described with reference to the drawings. Note that the specification definition table 201 and the resource definition table 401 for each request type are defined in advance.

  FIG. 6 is a flowchart showing the operation (request acceptance) of the request signal processing apparatus 101.

  When the request signal receiving unit 102 receives the request signal 106, the request signal receiving unit 102 determines whether or not the request signal processing apparatus 101 as a whole can perform request reception by performing a request reception determination process according to the flowchart illustrated in FIG. 6. The specific procedure is as follows.

  The request signal receiving unit 102 extracts the request type from the request signal 106 received from the outside (S601).

  The request signal reception unit 102 acquires the maximum request allowable number (whole apparatus) MEMRS from the request type execution table 201 corresponding to the extracted request type (S602).

  The request signal reception unit 102 acquires the number of execution requests for each request type of the entire apparatus corresponding to the corresponding request type from the entire apparatus execution state table 301 (hereinafter, the acquired number of execution requests for each request type is referred to as “meprs”). (S603).

  The request signal accepting unit 102 compares the MEPRS obtained by the process of step S602 described above with the magnitude of meprs obtained by the process of step S603 described above (S604).

  If meprs is less than MEPRS, the request signal receiving unit 102 determines “acceptable” (S605). If meprs is greater than or equal to MEPRS, the request signal accepting unit 102 determines that “acceptance is impossible” (S606).

  Next, the request signal 106 determined to be “acceptable” in the process of step S <b> 605 is delivered to the load distribution function unit 103. Then, the load distribution function unit 103 determines the resource (request process execution unit 104) to be actually allocated for the process related to the request signal 106 by performing the load distribution process.

  FIG. 7 is a flowchart showing resource distribution processing executed by the load distribution function unit 103.

  The load distribution function unit 103 extracts a request type from the received request signal 106 (S701).

  The load distribution function unit 103 searches and lists all the resources defined to be fixedly assigned the corresponding request type from the resource definition table 401 (S702).

  In the process of step S702 described above, the load distribution function unit 103 performs the first list determination process for the corresponding list when even one corresponding resource exists (S703).

  FIG. 8 is a flowchart showing an operation (first list determination process) of the request signal processing apparatus.

  The load distribution function unit 103 refers to the per-resource execution state table 501, and all the resources in which the number of requests being executed for each request type of the corresponding request type is equal to or greater than the maximum allowable request number (for each resource) MERPR are all extracted from the extracted resource list. Exclude (S801).

  The load distribution function unit 103 determines whether resources remain in the resource list excluded in the process of step S801 described above (S802).

  If at least one resource remains in the resource list, the load distribution function unit 103 determines “selectable” and returns to the main process together with the edited resource list (S803). On the other hand, if no resource remains in the resource list, the load distribution function unit 103 determines that “selection is not possible” and returns to the main process (S804).

  The load distribution function unit 103 determines the processing result of step S703 (first list determination processing) (S704).

  If no corresponding resource exists in the resource list in the process of step S702 described above, or if it is determined in the process of step S704 that "selection not possible" is returned, the load balancing function The unit 103 extracts a list of resources that are defined to be arbitrarily allocated from the resource definition table 401 and that has a number of requests being executed for each request type of the corresponding request type in the corresponding resource execution status table 501 that is 1 or more. (S705).

  If even one resource exists, the load distribution function unit 103 performs the first list determination process for the corresponding list (S706). Since the process of step S706 is the same as the process of step S703, the description thereof is omitted.

  The load distribution function unit 103 determines the processing result of step S706 (first list determination processing) (S707).

  If no corresponding resource exists in the process of step S705 described above, or if it is determined that “selection not possible” is returned in the process of step S707, the load distribution function unit 103 Again, for resources defined to be arbitrarily allocated from the resource definition table 401, resources with the number of requests being executed for each request type of all the request types in the per-resource execution status table 501 are extracted and listed (S708). .

  If even one corresponding resource exists, the load distribution function unit 103 performs the first list determination process for the corresponding list (S709). Since the process of step S709 is the same as the process of step S703 and step S706, the description thereof is omitted.

  The load distribution function unit 103 determines the processing result of step S709 (first list determination processing) (S710).

  If no corresponding resource exists in the process of step S708 described above, or if it is determined that “selection not possible” is returned in the process of step S710, the load distribution function unit 103 The remaining resources that are excluded from the search targets in the processes of step S702, step S705, and step S708 are listed (S711).

  If even one resource exists, the load distribution function unit 103 performs the second list determination process for the corresponding list (S712).

  FIG. 9 is a flowchart showing the operation (second list determination process) of the request signal processing apparatus.

  From the extracted resource list, the load distribution function unit 103 extracts a request type having a number of requests being executed for each request type of 1 or more from the execution status table 501 for each resource, and sums all the usage rates (Σ (allocated) Exclude all resources for which the sum of the number of requests being executed for each request type / MERPR of the corresponding request type) and the usage rate for one received request type (1 / MERPR of the additional request type) exceeds 100% ( S901).

  The load distribution function unit 103 determines whether or not there is a resource in the resource list that has been subjected to the processing in step S901 described above (S902).

  If at least one resource remains, the load distribution function unit 103 determines that “selection is possible”, and returns to the main process together with the edited resource list (S903). If there is no resource in the resource list, the load distribution function unit 103 determines that “selection is not possible” and returns to the main process (S904).

  The load distribution function unit 103 determines the processing result of the second list determination process in step S712 (S713).

  If no corresponding resource exists in the process of step S711 described above, or if it is determined that “not selectable” is returned in the process of step S713 described above, the load balancing function unit 103 Determines that the request cannot be accepted and ends the processing (S714).

  When it is determined as “selectable” in the determination result (S704, S707, S710, S713) of the first list determination process or the second list determination process described above, the load distribution function unit 103 starts the list determination process. An allocated resource is selected from the output resource list (S715).

  Note that the method for selecting an allocation resource from the resource list in the process of step S715 described above is not particularly limited in this embodiment, but, for example, a ratio method, a least connection method, and the like can be considered.

  The load distribution function unit 103 obtains the number of requests being executed for each request type of the corresponding request type in the entire apparatus execution status table 301 and the number of requests being executed for each request type of the request type related to the corresponding resource in the resource execution status table 501. Addition is performed (S716).

  Note that the resource (request processing execution unit 104) to which the execution of processing is assigned by the load balancing function unit 103 is a request being executed for each request type of the corresponding request type in the entire apparatus execution state table 301 after the corresponding request processing is completed. The number of requests being executed for each request type of the request type related to the resource in the execution status table 501 for each resource is subtracted.

(A-3) Effects of First Embodiment According to this embodiment, the following effects can be achieved.

  In the request signal processing apparatus 101, it is defined in advance in the resource definition table 401 whether a specific request type is assigned to each resource or an arbitrary request type can be assigned, and the load distribution function unit 103 defines the corresponding definition. By preferentially allocating resources that are already executing only the corresponding request among resources that are “fixed allocation” or resources that are “arbitrary allocation” (in other words, the same resource as much as possible) Allocation of functions), high execution efficiency can be obtained, and optimal load distribution can be performed when the processing load and the allocation allowable number per resource are different for each function.

  Also, for resources other than fixed allocation, multiple request types can be used together, so the number of resources is always greater than the number of request types when a fixed resource is selected according to the request type. In a small number of cases, it is possible to solve the problem that only a specific request type is preferentially assigned and other subsequent request types cannot secure resources.

(B) Other Embodiments In addition to the above-described embodiments, the following modified embodiments can be exemplified.

  (B-1) In the above embodiment, in the load distribution process of the request signal processing apparatus 101, the example of executing the process of step S705 and the process of step S708 has been shown, but either of these processes is omitted or both Without the processing, the request signal processing apparatus 101 may execute load distribution processing.

  DESCRIPTION OF SYMBOLS 101 ... Request signal processing apparatus, 102 ... Request signal reception part, 103 ... Load distribution function part, 104 ... Request processing execution part, 105 ... Memory | storage part, 106 ... Request signal, 201 ... Specification definition table for every request classification, 301 ... Overall apparatus execution status table 401... Resource definition table 501.

Claims (9)

A load distribution device connected to a plurality of request processing execution units that perform processing based on a request type included in a received request signal, and distributes the processing to any of the request processing execution units,
Information indicating the maximum number of requests that can be processed by the plurality of request processing execution units for each request type, information indicating the maximum number of requests that can be processed individually by the request processing execution unit for each request type, and a specific request type A storage unit that stores information on the request processing execution unit that preferentially distributes the request processing based on the information, and information indicating the number of requests being processed by the request processing execution unit for each request type;
A distribution destination determination unit that determines a distribution destination in one of the plurality of request processing execution units based on the request type included in the received request signal and the information stored in the storage unit. A load balancer. If there is the request processing execution unit that preferentially distributes processing corresponding to the request type of the received request signal, the distribution destination determination unit lists the request processing execution unit that exists. The load distribution apparatus according to claim 1, wherein the request processing execution unit as a distribution destination is determined from an execution unit candidate list. When the allocation destination determination unit determines that the request processing execution unit as the allocation destination does not exist in the request processing execution unit candidate list, the allocation destination determination unit performs processing corresponding to the request type included in the request signal as 1 The load according to claim 2, wherein the request processing execution unit that has been executed is re-listed, and the request processing execution unit is determined from the re-listed request processing execution unit candidate list. Distributed device. The distribution destination determination unit executes processing corresponding to the request type included in the request signal when it is determined that the request processing execution unit as the distribution destination does not exist in the request processing execution unit candidate list 4. The load according to claim 2, wherein the request processing execution unit that has not been re-listed is re-listed, and the request processing execution unit is determined from the re-listed request processing execution unit candidate list. Distributed device. If the allocation destination determination unit determines that the request processing execution unit as the allocation destination does not exist in the request processing execution unit candidate list, the remaining request processing execution that has never been listed is performed. 5. The load distribution apparatus according to claim 2, wherein the request processing execution unit is determined from the re-listed request processing execution unit candidate list. In the request processing execution unit, the allocation destination determination unit processes the number of requests being executed by the request processing execution unit corresponding to the request type included in the request signal from the request processing execution unit candidate list. The load according to any one of claims 2 to 5, wherein when the number of requests is more than the maximum possible number, the corresponding request processing execution unit is excluded from the request processing execution unit candidate list. Distributed device. The allocation destination determination unit is configured to calculate, from the request processing execution unit candidate list, the number of requests being processed for each request type by the request processing execution unit with the maximum number of requests that can be processed for each corresponding request type. The request processing execution unit in which the sum of the sum of the divided values and the reciprocal of the number of requests that can be processed at the maximum for the request type corresponding to the request type included in the request signal is excluded. Item 7. The load balancer according to Item 6. A computer that is connected to a plurality of request processing execution units that perform processing based on the request type included in the received request signal, and is mounted on a load distribution device that distributes the processing to any of the request processing execution units,
Information indicating the maximum number of requests that can be processed by the plurality of request processing execution units for each request type, information indicating the maximum number of requests that can be processed individually by the request processing execution unit for each request type, and a specific request type A storage unit that stores information on the request processing execution unit that preferentially distributes the request processing based on the information, and information indicating the number of requests being processed by the request processing execution unit for each request type;
Functions as a distribution destination determination unit that determines a distribution destination to one of the plurality of request processing execution units based on the request type included in the received request signal and the information stored in the storage unit A load balancing program characterized by A processing apparatus comprising the load distribution apparatus according to any one of claims 1 to 7,
A request signal receiving unit for determining whether or not the received request signal can be received;
A processing apparatus comprising: a plurality of request processing execution units that perform processing based on a request type included in the request signal.

Images