JP6922670B2 - Resource determination device, resource determination method and resource determination processing program - Google Patents

Description

Embodiments of the present invention relate to a resource determination device, a resource determination method, and a resource determination processing program.

Generally, when a telecommunications carrier provides a communication service (hereinafter, may be referred to as a service) to a customer, a communication path (hereinafter, sometimes referred to as a service) that receives a request from the customer for a service layer and forwards traffic in response to the request. By assigning a link resource (hereinafter referred to as a link resource) and a node (hereinafter referred to as a functional resource) in which a network function for processing traffic is arranged, the network can be made available. Therefore, in the management of network services, how to combine resources and determine parameters according to the demands of the service layer are important factors. In general, these decisions depend on the skill of the operator operator.

When providing communication services in a legacy network, the degree of freedom in determining resource combinations and parameters is low, so operators can acquire the skills to provide communication services in a short time.

Virtualization network technology NFV (Network Functions Virtualization) can generate various network functions such as Firewall (FW) function and Web Server (WS) function on a general-purpose server, and the amount of resources allocated to each function. For example, the processing capacity of resources can be freely changed by freely changing the number of CPUs and the amount of memory. In the B2B2X (Business to Business to Consumer) era, service providers (middle B, users) end users by using flexible and inexpensive communication services built by NFV technology provided by telecommunications carriers (First B). It is expected that the number of forms of providing services to (X) will increase. The requirements for the telecommunications carrier for this situation are the provision of telecommunications services that meet the demands of the service layer from the telecommunications carriers and the efficient resource management for realizing the telecommunications services.

A specific example of the service layer request from the service provider will be described. When applying for a communication service, a service provider that provides online banking is interested in the online banking service, which is a function of the service, the safety of the communication service, and the number of users that the communication service can accommodate. Service providers that provide video services are paying attention to the functions of the services, such as video services, service processing capacity, and bandwidth. Telecommunications carriers are required to provide services in response to the demands of these service layers.

Efficient resource management for realizing the communication service is to determine an appropriate resource type and resource parameter that constitute the service according to the request of the service layer from the service provider. As a specific example, a combination of virtual network resources such as deep packet inspection (DPI), firewall, web server, application server (AS), etc. is determined according to the service layer requirement of online banking. Determining the value of a parameter representing the processing capacity of each resource is resource management according to the request of the service layer.

With the diversity of service layer requirements and the high degree of freedom in setting virtualization resources, operators of telecommunications carriers have advanced resource management skills in order to efficiently manage resources according to service layer requirements. Is required. This raises the issue of significantly increasing the complexity of service provision and management costs in virtualized networks.

For example, when a communication service user (service provider) applies for a virtualized network service using a request at the service layer, the user expects the communication provider to manage resources. In this situation, when a carrier provides a virtualized network service to a user, the carrier operator is required to have a high degree of skill in order to determine resources according to the user's service layer requirements. NS.

The request of the service layer includes a request regarding the function of the service and the load of the service. Service functions include, for example, web service, CDN (Content Delivery Network) service, and high security, and requests for service load include, for example, Work load 100,000 users and 1000 request / second.

In addition, the contents of the resource layer to be determined include the combination of resources for realizing the service and the value of the corresponding resource parameter. As for the combination of resources for realizing the service, for example, in order to realize the web service, three types of resources consisting of firewall, web server, and data base are used.
In addition, the value of the resource parameter includes, for example, the location of the instance of the resource, the number of CPUs of the instance of the corresponding resource, and the amount of memory.

Next, the prior art ETSI ENI will be described.
ETSI ENI ISG was established in February 2017 as a working group under the control of ETSI (European Telecommunications Standards Institute) to study the application of AI technology to the field of operations. The aim of ETSI ENI ISG is to utilize AI technology and context-aware policies to tailor the communication services provided and assist operators according to the individual needs of users, service provision conditions and business goals. That is.

ENI Engine is an architecture proposed by ENI (Experiential Network Intelligence) to improve the operator experience using Network Intelligence (see, for example, Non-Patent Document 1). The ENI Engine mainly consists of the following five functional blocks. Each functional block is at the concept level at the time of writing Non-Patent Document 1.

(1) Intelligent Service Deployment
(2) Intelligent Policy Control
(3) Intelligent Resource Management
(4) Intelligent Monitoring
(5) Intelligent Analyzing and Prediction
Of these, ENI's efforts will be described for (1) Intelligent Service Deployment, which is the functional block most closely related to the present invention.

In this architecture, the customer service requester is input to the system, and the service template function (Service Template), language transformation function (Language Transformation), and service combination function (Service) are input through the Service Intent API. Composition and Management) are called respectively.
Using the information generated by these functions, instructions are given to the Network Controller through the Service Delivery API, and the Network Controller controls the resources (Network Element) through the Resource Management API, and the service is provided.

However, in this architecture, the concrete expression method, analysis method, resource combination determination method, and system processing of Customer Service Requester are not clearly explained.

Open Day Light (eg Open Day Light NIC
See <https://wiki.opendaylight.org/view/Network_Intent_Composition:Main>) and ONOS (Open Network Operating System) (for example, ONOS Intent Framework).
Software Defined Network (SDN) study groups such as <https://wiki.onosproject.org/display/ONOS/Intent+Framework>) use Intent to hide the complexity of virtualized network management. I am considering IBNM to control the network. The aim of this organization is to control the network by asking the operator for "What I want" instead of "Do it like this".

FIG. 30 is a diagram illustrating an example of IBNM (Intent-based Network Management).
IBNM will be explained using the example of Network Intent Composition (NIC) of Open Day Light. Here, a Headquarter is provided between Branch 1 and Branch 2, a link c1 having a bandwidth of 10 Gbps is determined between Branch 1 and Headquarter, the flow f1 is passed through the Firewall (FW) of the Headquarter, and the bandwidth of the link c1 is determined. Is an example of controlling 10 Gbps in the daytime and 1 Gbps in the nighttime.

Conventionally, it is necessary to determine complicated resources such as determining parameters such as the IP of the link resource and transfer rules at each switch in the route. With IBNM, the above control can be realized simply by describing the requirements to be realized, and the complexity of determining link parameters other than the control to be realized can be hidden. The requirements are described, for example, "Connection c1 Endnodes (branch1, HQ) Properties 10G" (corresponding to "(A) c1: 10G" in Fig. 30), "Policy ID1 Apply to f1 action go through (firewall, WoC)". (Corresponding to "(B) f1: Add FW" in FIG. 30).

On the other hand, applying the existing IBNM to service management in a virtualized network has the following two problems.
The first issue is that the management target of IBNM is limited to link resource management, and it is not possible to handle the management of functional resources.
The second problem is that the function of combining appropriate resources according to the request of the service layer and the function of determining the parameters of the functional resource according to the request of the service layer have not been studied.

Raymond Forbes, "Experimental Networked Intelligence", <https://portal.etsi.org/Portals/0/TBpages/ENI/Docs/ISG_ENI_presenatation.pdf> Customizing the tokenizer spaCy Usage Documentation, <https://spacy.io/docs/usage/customizing-tokenizer> Part-of-speech tagging spaCy Usage Documentation, <https://spacy.io/docs/usage/pos-tagging> Rule-based matching spaCy Usage Documentation, <https://spacy.io/docs/usage/rule-based-matching>

The method of determining the resource from the request of the service layer in the current situation will be described.
Here, the first method (Cloud Consultant) and the second method (Self-management) will be described.

In the first method (Cloud Consultant method), the sales representative of the telecommunications carrier grasps the request of the service provider as the request of the service layer through hearing the request of the service layer, and the operator of the telecommunications carrier responds to this request. Determine resources accordingly. This method can express the demands of the service layer, but it requires the operation of the sales staff of the telecommunications carrier and the operator, and the cost for realizing it is high.

The second method (Self-management) is a method in which resource management is entrusted to a service provider. In this method, for example, the service provider itself converts the request of the service layer into the request for the resource, and the required resource is expressed through the resource management interface Dashboard provided by the telecommunications carrier. However, this method requires the service provider to have knowledge of the virtual network, and the service provider cannot directly express the request of the service layer.

An object of the present invention is to provide a resource determination device, a resource determination method, and a resource determination processing program capable of determining a resource combination and a resource parameter suitable for a service layer request.

In order to achieve the above object, the first aspect of the resource determination device in one embodiment of the present invention is a service layer in which the resource determination device is a request expressed by a sentence described in natural language for a communication service. The analysis means for analyzing the request of the above, the combination determination means for determining the combination of resources suitable for the request of the service layer analyzed by the analysis means, and the resource included in the combination of the resources determined by the combination determination means. The analysis means includes a parameter determining means for determining a parameter , the analysis means decomposes a sentence indicated by a request of the service layer into words, and for each of the decomposed words, a tag corresponding to the attribute of the word is attached. By applying an analysis rule according to the attribute to each attribute of the word to which the tag is attached, at least one request of the service layer is analyzed .

Customizing a second aspect of the resource determination device configured as described above, in the first aspect, said combination determining means, with respect to the basic configuration of a combination of the resources, in accordance with the requirements of service layer analyzed by the analyzing means -By applying a rule and customizing the basic configuration, the combination of the resources is determined.

In the third aspect of the resource determination device having the above configuration, in the first aspect, the parameter determination means affects the parameters of the resources included in the combination of resources determined by the combination determination means. The relationship between the above is learned to generate a model, and the generated model is used to determine the parameters of the resources determined by the combination determination means.
A fourth aspect of the resource determination device having the above configuration is, in the first aspect, the analysis means analyzes a request of a service layer expressed in a natural language created according to a defined expression rule for a communication service. It is something like that.

One aspect of the resource determination method performed by the resource determination device in one embodiment of the present invention is to analyze a request of a service layer, which is a request expressed by a sentence described in a natural language, for a communication service, and the analysis is performed. Determining the combination of resources suitable for the requested service layer, determining the parameters of the resources included in the determined combination of resources, and analyzing the word the sentence indicated by the requested service layer. It is decomposed into By applying the rule, it includes analyzing at least one request of the service layer .

One aspect of the resource determination processing program according to the embodiment of the present invention is to make the processor function as each of the means of the resource determination device in any one of the first to fourth aspects.

According to the first aspect of the resource determination device according to the first embodiment of the present invention, the combination of resources and the resource parameters can be determined without depending on the operator, so that the human cost can be reduced.

Further , according to the first aspect of the resource determination device according to the first embodiment of the present invention, the request of the service layer expressed in natural language can be appropriately analyzed.

According to the second aspect of the resource determination device according to the second embodiment of the present invention, it is possible to determine the resource configuration suitable for the request of the service layer.

According to the third aspect of the resource determination device according to the third embodiment of the present invention, it is possible to determine the resource parameters affected by a plurality of types of elements without depending on the operator.

That is, according to the present invention, it is possible to determine a resource combination and a resource parameter suitable for the service layer requirements.

The figure which shows an example of the outline of the combination of resource determination and the determination of a resource parameter in one embodiment of the present invention. The figure which shows the application example which concerns on the analysis of the request of the service layer of the resource determination apparatus in one Embodiment of this invention. The figure explaining an example of the expression rule of the request of the service layer applied to the resource determination apparatus in one Embodiment of this invention. The figure explaining an example of the expression rule of the request of the service layer applied to the resource determination apparatus in one Embodiment of this invention. The figure which shows an example of the expression of the request of the service layer according to the expression rule by the resource determination device in one Embodiment of this invention. The figure which shows an example of the outline of the analysis of the request of a service layer by the service layer request analysis part of the resource determination apparatus in one Embodiment of this invention. FIG. 5 is a flowchart showing an example of a procedure for analyzing a service layer request by the service layer request analysis unit of the resource determination device according to the embodiment of the present invention. The figure explaining an example of Tokenize by the service layer request analysis part of the resource determination apparatus in one Embodiment of this invention. The figure explaining an example of POS (Part of Speech Tagging) by the service layer request analysis part of the resource determination apparatus in one Embodiment of this invention. The figure explaining an example of Rule based Matching by the service layer request analysis part of the resource determination apparatus in one Embodiment of this invention. The figure which shows the application example which concerns on the determination of the resource combination of the resource determination apparatus in one Embodiment of this invention. The flowchart which shows an example of the procedure of the resource combination determination by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the determination of the resource combination by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining an example of the basic combination repository of services used by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure which shows an example of the customization rule used by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure which shows an example of the format of the customization rule used by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure which shows an example of the format of the customization rule used by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure which shows an example of the format of the customization rule used by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The flowchart which shows an example of the processing procedure concerning the customization engine by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the 1st example of the resource combination determined by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the 2nd example of the resource combination determined by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the 3rd example of the resource combination determined by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the 4th example of the resource combination determined by the resource combination determination part of the resource determination apparatus in one Embodiment of this invention. The figure which shows the application example concerning the determination of the resource parameter of the resource determination apparatus in one Embodiment of this invention. The figure which shows an example of the outline of the processing by the resource parameter determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the learning phase and the operation phase in the processing by the resource parameter determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the specific example of the learning phase in the processing by the resource parameter determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining the specific example of the operation phase in the processing by the resource parameter determination part of the resource determination apparatus in one Embodiment of this invention. The figure explaining an example of the effect which occurs by using the resource determination apparatus in one Embodiment of this invention. The figure explaining an example of IBNM (Intent-based Network Management).

Hereinafter, an embodiment according to the present invention will be described with reference to the drawings.
FIG. 1 is a diagram showing an example of an outline of resource combinations and resource parameter determination in one embodiment of the present invention. In this embodiment, the service provider expresses the request of the service layer as a service requester.
Further, according to the received Service Requester, the resource determination device that realizes one embodiment of the present invention determines an appropriate combination of resources and resource parameters as a Resource Descriptor.

Next, the expression and analysis of the request of the service layer will be described. FIG. 2 is a diagram showing an application example of the resource determination device according to the embodiment of the present invention for analyzing the demands of the service layer.
As shown in FIG. 2, the resource determination device 10 includes a service layer request analysis unit 11, a resource combination determination unit 12, and a resource parameter determination unit 13. The service layer request analysis unit 11, the resource combination determination unit 12, and the resource parameter determination unit 13 of the resource determination device 10 can be realized by a device as a computer device such as a personal computer (PC). For example, a computer device includes a processor such as a CPU (Central Processing Unit), a volatile memory connected to the processor, a non-volatile memory, and a communication interface.

The service layer request analysis unit 11 inputs the service requester and outputs the analysis result of the service requester to the resource combination determination unit 12 and the resource parameter determination unit 13.
The resource combination determination unit 12 determines the resource combination according to the analysis result and outputs the resource combination.
The resource parameter determination unit 13 determines the resource parameter according to the output result and outputs the resource parameter.
The output by the resource combination determination unit 12 and the output by the resource parameter determination unit 13 are output to the virtualization resource management system as a resource descriptor, respectively.

Next, the analysis of the service layer request related to the "Input: Service Requester" in the frame of W1 shown in FIG. 2 and the service layer request analysis unit 11 will be described.
In this embodiment, the Controlled Natural Language ( CNL: Limited natural language) describes how to express service layer requirements.

CNL is a natural language created according to pre-defined expression rules. Compared to fully natural language, this CNL has the advantages of less ambiguity, more accurate representation of requirements, easier analysis of requirements, and easier loading into the system. Therefore, it is possible to reduce the operation of the sales staff of the telecommunications carrier and accurately express the demands of the service layer.

Here, in order to express the request of the service layer, the expression rule of the request of the service layer is defined in the CNL format shown below.
3 and 4 are diagrams showing an example of a service layer request expression rule applied to the resource determination device according to the embodiment of the present invention.
As shown in FIG. 3, the service layer request expression rule is represented by "<ACTION><SERVICE><CONJ><DESCRIPTOR>".

<ACTION> represents an operation on the service.
<SERVICE> represents the type of service you are trying to open or an identifier that represents an existing service.
<CONJ> connects <SERVICE> and <DESCRIPTOR>.

As shown in FIG. 3, there are a plurality of types of words that can be used for <ACTION>, <SERVICE>, and <CONJ>.
<DESCRIPTOR> represents the attribute (requirement) of the service required by <SERVICE>. In the example shown in FIG. 4, there are five types of service attributes, WORKLOAD, LINK PROPERTY, REDUNDANCY, SECURITY, and REGION, and the grammar of the expression rule is associated with each type. In FIG. 4, words that can be used for tags included in the expression rule are shown.

FIG. 5 is a diagram showing an example of expression of a request of a service layer according to an expression rule by a resource determination device according to an embodiment of the present invention.
According to the expression rule of the request of the service layer, the request of the service layer can be expressed by Examples A, B, C and the like shown in FIG.

Next, an example of analyzing the request of the service layer expressed as in Example A above will be described. FIG. 6 is a diagram showing an example of an outline of analysis of a service layer request by the service layer request analysis unit of the resource determination device according to the embodiment of the present invention.
When the requirements analyzed for the service layer requirements shown in Example A are expressed in a .json (JavaScript (registered trademark) Object Notation) file, "SECURITY" is "high security" as shown in FIG. The analysis result is that "WORKLOAD" is "1000 request / second" and "SERVICE TYPE" is "web service".

Next, each procedure of analyzing the request of the service layer will be described. This procedure is Tokenize, Part of Speech Tagging (POS), and Rule based Matching.
FIG. 7 is a diagram showing an example of a procedure for analyzing a service layer request by the service layer request analysis unit of the resource determination device according to the embodiment of the present invention. FIG. 8 is a diagram illustrating an example of Tokenize by the service layer request analysis unit of the resource determination device according to the embodiment of the present invention.
First, the service layer request analysis unit 11 performs Tokenize for the service layer request (S11) . Tokenize is an existing function that breaks down sentences into words.
Here, tokenize disclosed in Non-Patent Document 2 is used.

When Tokenize is performed for the service layer request shown in Example A above, the result of Tokenize is
You get {Start, web, service, that, handles, 1000, request / second, with, high, security}.

FIG. 9 is a diagram illustrating an example of POS by the service layer request analysis unit of the resource determination device according to the embodiment of the present invention.
The service layer request analysis unit 11 performs POS for the result of Tokenize (S12).

POS is an existing technology that attaches tags to words according to the content of the words. Here, POS tagging disclosed in Non-Patent Document 3 is used.

When POS tagging is performed on the result of Tokenize shown in FIG. 8, the result of POS tagging is obtained.
{Start: VERB, web: NOUN, service: NOUN, that: CONJ, handles: VERB,
1000: NUM, request / second: NOUN, with: CONJ, high: ADJ, security: NOUN} is obtained.
The tags given here are VERB (verb), NOUN (noun), CONJ (conjunction), NUM (number), ADJ (adjective), and ADV (adverb).

FIG. 10 is a diagram illustrating Rule based Matching by the service layer request analysis unit of the resource determination device according to the embodiment of the present invention.
The service layer request analysis unit 11 performs Rule based Matching for the POS result (S13).

Rule-based matching is an existing technique that identifies and classifies the attributes of one or more words according to predefined analysis rules. Here, the rule-based matching disclosed in Non-Patent Document 4 was used. As a result of rule-based matching, requests for each type of service attribute are categorized and output. Here, the requirements for each type of service attribute are called atom_requirements. atom_requirements is the input of the following resource combination determination function and the resource parameter determination function.

When Rule based Matching is performed on the result of POS tagging shown in FIG. 9, the result of Rule based Matching is obtained.
You get {"SECURITY": "high security", "WORKLOAD": "1000 request / second", "SERVICE TYPE": "web service"}. This result is the final analysis result of the service layer request.

Here, the analysis rules used in Rule based Matching will be described.
This analysis rule is designed corresponding to the basic expression rule as shown in FIG. The analysis rule designed corresponding to the basic expression rule as shown in FIG. 4 includes the analysis rule of WORKLOAD, the analysis rule of LINK PROPERTY, and the analysis rule of SECURITY. An example of the logic of various analysis rules is shown below.

・ WORKLOAD analysis rule:
If input = <NUM> + users per second / transactions per second / request per second:
mark as WORKLOAD requirement;
・ Analysis rule of LINK PROPERTY:
If input = <ADJ or NUM> + GB / bandwidth / packet loss rate / priority:
mark as LINK PROPERTY requirement;
・ SECURITY analysis rules:
If input = <ADJ> / <ADV> + security / protection:
mark as SECURITY requirement;

Next, the determination of the resource combination will be described. FIG. 11 is a diagram showing an application example of the resource determination device according to the embodiment of the present invention for determining a resource combination. Next, the determination of the resource combination related to the “Output: resource combination” in the frame of W2 shown in FIG. 11 and the resource combination determination unit 12 will be described.
After S11, the resource combination determination unit 12 determines the resource combination according to the request of the service layer analyzed by the service layer request analysis unit 11 (S12).

Currently, determining a combination of resources according to the needs of the service layer generally depends on the skill of the operator. In addition, a resource combination template is created for each service, and the combination is determined with reference to this template. However, as the variation of services increases, it is necessary to create a large number of templates, the amount of data becomes enormous, labor is required, and the service provision time becomes long.

In the present embodiment, it will be described that a suitable resource combination is determined for each service layer request by using the basic service combination and the customization rule according to the analyzed service layer request.

FIG. 12 is a flowchart showing an example of a procedure for determining a resource combination by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
When determining the service configuration (resource configuration), the resource combination determination unit 12 first reads the basic resource configuration (basic combination) from the basic configuration repository according to the service type (S21). The repository including the basic configuration repository is realized by storing the basic configuration information in the non-volatile memory of the resource determination device 10. Next, the resource combination determination unit 12 reads out the request of the service layer (particularly, the redundancy requirement, the security requirement, and the regional requirement (in the case of CDN)) based on this basic configuration (S22), and the request of this service layer. The resource configuration is customized by adding and changing functions for the basic service combination based on (S23).

By automating the determination of the combination of resources, the operator does not have to spend much effort as in the past, so that the effect of reducing the human cost and the service provision time can be expected.

Further, instead of creating a resource combination template for each service, a small number of resource combination templates can be prepared, and resource combinations of various services can be generated through a customization engine.

The resource combination determination unit 12 is based on the basic service combination, applies customization rules according to the request of the service layer analyzed in S11, adds a function to the basic service combination, and changes the function for each request. Automatically generate a suitable resource combination (.yaml file).

FIG. 13 is a diagram illustrating an example of determination of a resource combination by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
In the example shown in FIG. 13, the types of functional resources are FW (Firewall), WS (Web Server), AS (Application Server), CS (Content Server), and LB (Load Balancer), which are the basics of the Web Service system. It is divided into a configuration and a basic configuration of a CDN (Content Delivery Network) system.

In the example shown in FIG. 13, the basic configuration of the Web Service system in the basic service combination includes the link resource Link that connects FW (Firewall), WS (Web Server), AS (Application Server) and these functional resources. be. In addition, the basic configuration of the CDN system in the basic service combination is a configuration that includes a link resource Link that connects CS (Content Server), LB (Load Balancer) and these functional resources.

FIG. 14 is a diagram illustrating an example of a basic combination repository of services used by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
The basic service combination is stored in the repository.

As shown in FIG. 14A, the basic configuration of the Web Service system is a configuration including the functional resources FW, WS, AS and the link resource Link that connects these functional resources, and is the basic configuration of the CDN system. The basic service combination in which is LB and CS can be defined as in the format shown in FIG. 14 (b). The image_URL in this format indicates the download destination address of the image of the corresponding resource.
Further, the tier shown in FIG. 14B shows the order of resources before and after in the flow of communication flow.

The resource combination determination unit 12 applies customization rules according to the request of the service layer based on the basic combination of services.
FIG. 15 is a diagram showing an example of a customization rule used by the resource combination determination unit of the resource determination device according to the embodiment of the present invention. As shown in FIG. 15, the format of the customization rule is ECA (Event Condition Action).

The Condition in the customization rule shown in FIG. 15 indicates the target service type of the customization rule. The Event in the customization rule shown in FIG. 15 represents the request of the service layer corresponding to the customization rule shown in FIG. This type of Event corresponds to the classification of service layer requests. The Action in the customization rule shown in FIG. 15 represents an operation on the service combination, for example, adding, changing, or deleting a resource.

16th, 17th, and 18th are diagrams showing an example of the format of the customization rule used by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
The customization rule for the "No. 1" column shown in FIG. 15 can be defined as shown in FIG. This customization rule defines that the Event is "high security", the Condition is "any service", and the Action is "add security DPI functionality to the basic configuration".

The customization rule for the "No. 2" column shown in FIG. 15 can be defined as shown in FIG. This customization rule defines that Event is "full redundancy", Condition is "any service", and Action is "addition and physical separation of the same configuration to the basic configuration".

The customization rule for the "No. 3" column shown in FIG. 15 can be defined as shown in FIG. In this customization rule, Event is "number of regions = n", Condition is "any service", and Action is "Add to n content servers (CS)". Defined.

FIG. 19 is a flowchart showing an example of a processing procedure related to the customization engine by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
First, the customization engine checks whether the SERVICE TYPE is included in the entered requirements (atom_requirements). (S31). This input requirement (atom_requirements) is the analysis result of the request of the service layer. If the entered requirements (atom_requirements) include the SERVICE TYPE, the corresponding basic service combination (resource composition) is fetched from the repository (S32), and the SERVICE TYPE is included. If not, it returns an error .
After S32, the customization engine extracts any customization rules corresponding to these requirements (S33, S34) according to the requirements other than the SERVICE TYPE included in the input requirements (S35). ..

Subsequently, the customization engine analyzes the priority and conflict of the extracted customization rule (S36), and returns an error if there is a contradiction. If there is no contradiction, the customization engine executes the customization rule in order of priority (S37, 38), and adds, modifies, or deletes resources to the basic resource combination according to the customization rule.
Finally, the customization engine generates a .yaml file based on the determined resource combination (S39).

FIG. 20 is a diagram illustrating a first example of a resource combination determined by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
In the example shown in FIG. 20, first, the resource combination determination unit 12 extracts the service basic configuration suitable for the service type from the repository. The extracted service type is a web service system, and the basic configuration of this web service system includes the functional resources FW, WS, DB and the link resource Link that connects the functional resources. The basic resource configuration is as follows.

Basic resource configuration: [FW] [Link] [WS] [Link] [DB]
Next, the resource combination determination unit 12 customizes the basic resource configuration according to the request of the service layer according to the customization rule based on the request of the service layer. Here, since the security requirement (SECURITY) in the customization rule is high security, the resource combination is determined by adding a security function called Deep Packet Inspection (DPI) to the basic resource configuration. As a result, the resource configuration is changed as follows.

Resource configuration after customization: [DPI] [Link] [FW] [Link] [WS] [Link] [DB]
As a result, the resource configuration suitable for the service layer request is determined.

FIG. 21 is a diagram illustrating a second example of a resource combination determined by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
In the example shown in FIG. 21, first, the resource combination determination unit 12 extracts the service basic configuration suitable for the service type from the repository. The extracted service type is a web service system, and the basic configuration of this web service system includes the functional resources FW, WS, DB and the link resource Link that connects the functional resources. The basic resource configuration is as follows.

Basic resource configuration: [FW] [Link] [WS] [Link] [DB]
Next, the resource combination determination unit 12 customizes the basic resource configuration according to the request of the service layer according to the customization rule based on the request of the service layer. Here, since the redundancy requirement (REDUNDANCY) in the customization rule is full redundancy, an exact same configuration as the above basic resource configuration is added, and this added configuration and the above basic resource configuration are physically combined. Separation determines the combination of resources. As a result, the resource configuration is changed as follows.

Resource configuration after customization:
[FW] [Link] [WS] [Link] [DB]
[FW] [Link] [WS] [Link] [DB]
As a result, the resource configuration suitable for the service layer request is determined.

FIG. 22 is a diagram illustrating a third example of a resource combination determined by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
In the example shown in FIG. 22, first, the resource combination determination unit 12 extracts the service basic configuration suitable for the service type from the repository. The service type in this extracted configuration is a CDN Service system, and the basic configuration of this CDN Service system includes the functional resources LB and CS and the link resource Link that connects the functional resources. The basic resource configuration is as follows.

Basic resource configuration: [LB] [Link] [CS]
Next, the resource combination determination unit 12 customizes the basic resource configuration according to the request of the service layer according to the customization rule based on the request of the service layer. Here, since the REGION requirement in the customization rule is "A city, B city, C city", the resource combination determination unit 12 sets the number of CSs in the basic resource configuration to three according to the customization rule. To. As a result, the resource configuration is changed to a configuration in which three [Link] [CS] are connected in parallel with respect to [LB].
As a result, the service configuration suitable for the request of the service layer is determined.

FIG. 23 is a diagram illustrating a third example of a resource combination determined by the resource combination determination unit of the resource determination device according to the embodiment of the present invention.
In the example shown in FIG. 23, first, the resource combination determination unit 12 extracts the service basic configuration suitable for the service type from the repository. The service type in this extracted configuration is a web service system, and the basic configuration of this web service system includes the functional resources FW, WS, DB and the link resource Link that connects the functional resources. The basic resource configuration is as follows.

Basic resource configuration: [FW] [Link] [WS] [Link] [DB]
Next, the resource combination determination unit 12 retrieves the customization rule from the repository based on the request of the service layer. Here, the extracted customization rules are two customization rules (SECURITY: high security and REDUNDANCY: full redundancy) corresponding to the requirements. The order in which these customization rules are executed changes the resource configuration determined after customization.

The resource combination determination unit 12 compares the priorities of the retrieved customization rules. In the example shown in FIG. 15, the customization rule priority for the SECURITY requirement is higher than the customization priority for the REDUNDANCY requirement. According to this priority, the resource combination determination unit 12 first executes the customization according to the customization rule related to the SECURITY requirement as the first customization, that is, adds the security function DPI to the basic resource configuration.
The resource configuration after executing this customization is as follows.

[DPI] [Link] [FW] [Link] [WS] [Link] [DB]
Subsequently, the resource combination determination unit 12 executes the second customization of the customization according to the customization rule related to the REDUNDANCY requirement, that is, adds the same configuration as the configuration to the configuration after the first customization is executed. do.

The resource configuration after executing the second customization is as follows.
[DPI] [Link] [FW] [Link] [WS] [Link] [DB]
[DPI] [Link] [FW] [Link] [WS] [Link] [DB]

Next, determination of resource parameters will be described. FIG. 24 is a diagram showing an application example of the resource determination device according to the embodiment of the present invention for determining resource parameters. Here, the determination of the resource parameters related to the "Output: resource parameter" in the frame of W3 shown in FIG. 24 and the resource parameter determination unit 13 will be described.

In general, the allocation of computational resources to each resource at the request of the service layer, that is, the determination of resource parameters, depends on the operator. However, many factors need to be considered in determining resource parameters, including service layer requirements. For example, it is necessary to determine the resource parameters in consideration of all factors such as the type of processing, the resource implementation method, the resource utilization target set by the cloud provider, and the required workload.

Therefore, in order to determine the resource parameters, the operator is required to have a high skill, which causes a human cost, and the level of the operator's skill causes inefficiency in resource allocation.

FIG. 25 is a diagram showing an example of an outline of processing by the resource parameter determination unit of the resource determination device according to the embodiment of the present invention.
In the present embodiment, as shown in FIG. 25, the resource parameter determination unit 13 determines the number of CPUs, which is a parameter of each resource (for example, DPI, FW, WS, DB) determined by the resource combination determination unit 12. Determine the memory size and disk size.

Determining these parameters is directly related to the workload requirements of the service (eg 1000 request / second, 1 million users). Also, the type of processing (eg FW, AS, WS, CS, Encoder, DPI), carrier utilization target (eg virtual resource utilization <80%), host server status (eg host server). CPU utilization) and hardware settings (for example, linking a virtual CPU and a physical CPU) affect the determination of resource parameters.

The resource parameter determination unit 13 models the relationship between the service workload requirement, the resource parameter, and other influential factors by learning (Linear Regression, DDN Regression), and determines the resource parameter using this model. ..

Next, the learning phase and the operation phase in the processing by the resource parameter determination unit 13 will be described. FIG. 26 is a diagram illustrating a learning phase and an operation phase in processing by the resource parameter determination unit of the resource determination device according to the embodiment of the present invention.
In the learning phase, a model for resource parameter determination is generated by performing machine learning on parameters determined in the past by the operator and known design / operation data (resource utilization rate of virtualized resource instances) generated by this determination. It is a phase.

The operation phase is a phase in which resource parameters are determined by inputting resource requirements, resource processing types, resource target utilization rates, etc. into the model generated in the learning phase.

Next, a specific example of the learning phase will be described. FIG. 27 is a diagram illustrating a specific example of a learning phase in processing by the resource parameter determination unit of the resource determination device according to the embodiment of the present invention.
In the learning phase, the resource parameter determination unit 13 uses machine learning linear regression analysis (Linear Regression) and DDN regression analysis (Deep Neural Network Regression) model (for example, Tensorflow Estimators. (Https://www.tensorflow.org/). (See extend / estimators)) to learn the resource parameters determined in the past by the operator, the above-mentioned influential factors (processing type, host server utilization rate, etc.), and operational data. An example of training data to be trained is as follows.

-Resource processing type (VNF type)
· Workload requirements
-Resources decided by operators
· Host utilization ratios
-Hardware conditions
-VNF utilization ratios of each resource instance when resources are executed in the above environment.
Next, a specific example of the operation phase will be described. FIG. 28 is a diagram illustrating a specific example of an operation phase in processing by the resource parameter determination unit of the resource determination device according to the embodiment of the present invention.
In the operation phase, the resource parameter determination unit 13 determines parameters according to the required workload requirements, the target resource utilization rate of each instance determined by the cloud provider itself, etc., using the model generated in the learning phase. .. The following is an example of the data to be input to the model.

Resource processing type (VNF type)
Required workload requirements
Host utilization ratios
Hardware conditions
Objective VNF use ratios
An example of the data output from the model is as follows.

Resource parameters (number of CPUs, memory size, disk size) (calculated parameters)
Finally, the resource parameter determination unit 13 adds a flavor attribute (for example, Openstack flavor) to the .yaml file generated by the resource combination determination unit 12 based on the determined parameters.
(See https://docs.openstack.org/horizon/latest/admin/manage-flavors.html). The flavor attribute is an attribute that represents the number of CPUs, memory size, and disk size of the resource.

Next, the effect produced by one embodiment of the present invention will be described. FIG. 29 is a diagram illustrating an example of the effect produced by using the resource determination device according to the embodiment of the present invention.

In the example shown in FIG. 29, when the determination of the resource combination and the determination of the resource parameters depend on the operator and the resource determination device in one embodiment of the present invention is not used, the required number of Cloud Specialists for the number of sold clouds is , More than the current number of Cloud Specialists, there will be a shortage of human resources.
In addition, there are many design costs for income in small and medium-sized enterprises and large enterprises that perform cloud computing. In addition, when the number of cloud sales exceeds a certain level, the average service provision time for the number of cloud sales increases significantly.

On the other hand, when the resource combination determination and the resource parameter determination do not depend on the operator and the resource determination device according to the embodiment of the present invention is used, the required number of Cloud Specialists for the number of sold clouds is the current number. It will be less than the number of Cloud Specialists, and the above-mentioned shortage of human resources will not occur. That is, the number of human resources required to design a virtualized network can be reduced, and the human resource development cost can be reduced.

In addition, the design cost for income in small and medium-sized enterprises and large enterprises that perform cloud computing is reduced as compared with the case of relying on the above operator.
That is, since the resource determination device in one embodiment of the present invention supports the service design, it is possible to reduce labor and time costs, secure profits, and increase profits.
Further, even if the number of cloud sales exceeds a certain level, the average service provision time for the number of cloud sales is shorter than that in the case of relying on the above operator.
That is, since the service design can be automated by the resource determination device according to the embodiment of the present invention, the service provision time can be shortened and the sales can be expanded.

The present invention is not limited to the above embodiment, and can be variously modified at the implementation stage without departing from the gist thereof. In addition, each embodiment may be carried out in combination as appropriate, in which case the combined effect can be obtained. Further, the above-described embodiment includes various inventions, and various inventions can be extracted by a combination selected from a plurality of disclosed constituent requirements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in the embodiment, if the problem can be solved and the effect is obtained, the configuration in which the constituent requirements are deleted can be extracted as an invention.

Further, the methods described in each embodiment include, for example, a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), an optical disk (CD-ROM,) as a program (software means) that can be executed by a computer (computer). It can be stored in a recording medium such as a DVD, MO, etc.), a semiconductor memory (ROM, RAM, flash memory, etc.), or transmitted and distributed by a communication medium. The program stored on the medium side also includes a setting program for configuring the software means (including not only the execution program but also the table and the data structure) to be executed by the computer in the computer. A computer that realizes this device reads a program recorded on a recording medium, constructs software means by a setting program in some cases, and executes the above-mentioned processing by controlling the operation by the software means. The recording medium referred to in the present specification is not limited to distribution, and includes a storage medium such as a magnetic disk or a semiconductor memory provided in a device connected inside a computer or via a network.

10 ... resource determination device, 11 ... service layer request analysis unit, 12 ... resource combination determination unit, 13 ... resource parameter determination unit.

Claims (6)

For communication services, analysis means for analyzing service layer requests, which are requests expressed in sentences written in natural language,
A combination determining means for determining a combination of resources suitable for the service layer request analyzed by the analysis means, and a combination determining means.
It is provided with a parameter determining means for determining the parameters of the resources included in the combination of resources determined by the combination determining means .
The analysis means
The sentence indicated by the request of the service layer is decomposed into words, and the sentence is decomposed into words.
Each of the decomposed words is given a tag according to the attribute of the word.
By applying an analysis rule according to the attribute to each attribute of the word to which the tag is attached, at least one request of the service layer is analyzed.
Resource determinant. The combination determining means
To the basic configuration of the combination of the resource, then apply the customization rule in accordance with the requirements of service layer which has been analyzed by said analysis means, by customizing the basic configuration, determines the combination of the resource,
The resource determination device according to claim 1. The parameter determining means is
A model is generated by learning the relationship between a plurality of types of elements that affect the resource parameters included in the resource combination determined by the combination determination means.
Using the generated model, the parameters of the resources determined by the combination determination means are determined.
The resource determination device according to claim 1. The analysis means
For communication services, analyze the demands of the service layer expressed in natural language created according to the specified expression rules.
The resource determination device according to claim 1. It is a resource determination method performed by the resource determination device.
For communication services, analyze the needs of the service layer expressed in natural language,
Determine the combination of resources that suits the requirements of the analyzed service layer.
Determining the parameters of the resources included in the combination of the determined resource,
The analysis is
Breaking down the sentence shown in the service layer request into words
To give each of the decomposed words a tag according to the attribute of the word,
It includes analyzing at least one request of the service layer by applying an analysis rule according to the attribute to each attribute of the word to which the tag is attached.
Resource determination method. A resource determination processing program that causes a processor to function as each of the means of the resource determination device according to any one of claims 1 to 4.

Images