Resource-aware monitoring device and cloud integrated operation system including the same

Abstract

The monitoring device monitors the operation of at least one virtual machine running on a physical server within allocated resources of a cloud integrated operation system, and includes: a transceiver module that collects operation information generated by the virtual machine; a monitoring module that monitors whether the virtual machine is operating non-ideally based on the operation information collected by the transceiver module; a model operation module that operates a monitoring learning model that assists the monitoring function of the monitoring module based on a request from the monitoring module; and a schedule module that determines the operation time of the monitoring learning model via a predetermined timing determination method so that the monitoring learning model can be operated within the allocated resources.

Description

[Technical Field]

[0001] The present invention relates to a monitoring device and a cloud integrated operating system, and more particularly to a monitoring device and a cloud integrated operating system that operate a cloud by monitoring whether virtual machines are operating normally. [Background technology]

[0002] In recent years, the cloud market has been steadily expanding, with many services being built on the cloud, and the scale of the cloud has also been exploding. Since it is practically difficult for an administrator to manage hundreds or thousands of virtual machines one by one, technologies for automated system operation and management are being developed. The key to automated operation is the realization of the functions required for cloud operation using learning models based on machine learning / deep learning.

[0003] Unlike existing methods that require the entire analysis algorithm to be set up, machine learning overcomes the limitations of existing computer algorithms by generating models tailored to the purpose based on given data. Due to its superior functionality and convenience, machine learning has been incorporated into various industrial fields and is now being developed and commercialized.

[0004] However, machine learning and deep learning require a lot of time and cost, such as collecting and labeling a large amount of data, and learning models that require a lot of resources are treated as the proprietary assets of each development company and are stored and used without being made public. As a result, learning models with the same purpose are produced and used in various places, resulting in a social problem of wasted resources.

[0005] Furthermore, as the number of learning models and the amount of analysis used to efficiently operate the cloud increase, the operational burden increases, resulting in problems such as the inability to efficiently manage the cloud. Summary of the Invention [Problem to be solved by the invention]

[0006] The present invention is intended to solve the above problems and provides a resource-aware monitoring device that improves system efficiency by utilizing advanced technology, and a cloud integrated operation system including the same. [Means for solving the problem]

[0007] A monitoring device according to one embodiment of the present invention is a monitoring device that monitors the operation of at least one virtual machine running on a physical server within allocated resources of a cloud integrated operation system, and includes: a transceiver module that collects operation information generated by the virtual machine; a monitoring module that monitors whether the virtual machine is operating non-ideally based on the operation information collected by the transceiver module; a model operation module that operates a supervised learning model that assists the monitoring function of the monitoring module based on a request from the monitoring module; and a schedule module that determines the operation time of the supervised learning model using a predetermined timing determination method so that the supervised learning model can be operated within the allocated resources.

[0008] In addition, the monitoring learning model includes a first monitoring learning model that predicts the load that will occur in the cloud integrated operation system during a predetermined future period, and the scheduling module can determine the operation time based on the load predicted by the first monitoring learning model.

[0009] The predetermined time point determination method may determine the operation time point when a load is generated at a predetermined rate or less based on the allocated resources.

[0010] The method for determining the predetermined time point may be a method for determining the operation time point at a time point when the least load is generated within the predetermined future period.

[0011] The system may further include a model selection module that selects the supervised learning model when a predetermined operating condition is satisfied, and the predetermined operating condition may be a condition in which the system is operated so that a load lower than a predetermined threshold is applied.

[0012] In addition, the learning model may include a first supervised learning model that predicts the load to be generated in the cloud integrated operation system during a predetermined future period, and the model selection module may change the predetermined threshold based on the load predicted by the first supervised learning model.

[0013] a model selection module that selects the supervised learning model to be operated by the model operation module when predetermined operating conditions are satisfied; and a management device that manages virtual machines operated on a physical server, a monitoring device that monitors operation of the virtual machines, and a learning model recommendation device that recommends a supervised learning model required for the monitoring device to realize a monitoring function. The monitoring device further includes a transmission / reception module that collects information generated by the virtual machines, a monitoring module that monitors whether the virtual machines are operating non-ideally based on the virtual machine information collected by the transmission / reception module, a model operation module that operates the supervised learning model to assist the monitoring function of the monitoring module at a request of the monitoring module, a schedule module that determines an operation time of the supervised learning model through a predetermined time determination method so that the supervised learning model can be operated within allocated resources, and a model selection module that selects the supervised learning model to be operated by the model operation module when predetermined operating conditions are satisfied. The model selection module requests the learning model recommendation device to recommend the supervised learning model when the predetermined operating conditions are satisfied. The learning model recommendation device can recommend a learning model when predetermined recommendation conditions are satisfied, which is a condition similar to the supervised learning model by a first similarity or higher, from among pre-stored stored learning models.

[0014] In addition, even if there is no learning model among the stored learning models that satisfies the predetermined recommendation condition, the learning model recommendation device can recommend a learning model that satisfies a predetermined first additional condition that determines whether the condition is satisfied based on a second similarity that is lower than the first similarity, or a predetermined second additional condition that determines whether the condition is satisfied based on a virtual learning model generated by combining multiple stored learning models.

[0015] Furthermore, the predetermined first additional condition may be a condition that the population of pre-stored stored learning models that are similar to the supervised learning model at a second similarity level or higher, similar to the supervised learning model at a level less than the first similarity level, and exist based on the supervised learning model at a level less than the second similarity level is less than a predetermined number.

[0016] A monitoring method according to one embodiment of the present invention is a monitoring method for monitoring the operation of at least one virtual machine running on a physical server within allocated resources of a cloud integrated operation system via a monitoring device, and includes the steps of: collecting operation information generated in the virtual machine by a transceiver module; monitoring whether the virtual machine is operating non-ideally based on the operation information by a monitoring module; operating a supervised learning model that assists the monitoring function of the monitoring module based on a request of the monitoring module by a model implementation module; and determining the operation time of the supervised learning model through a predetermined time point determination method by a scheduler module so that the supervised learning model can be operated within the allocated resources.

[0017] A learning model recommendation device according to one embodiment of the present invention is a learning model recommendation device that recommends a learning model calculated by machine learning or deep learning in response to a consumer request, and includes: a receiving model that receives the consumer's request and calculation conditions for a target learning model, which is the learning model requested by the consumer; a storage module that stores stored learning models, which are previously generated learning models, and the calculation conditions for the stored learning models; a similarity judgment module that judges the similarity between the target learning model and the stored learning model; and a recommendation module that recommends to the consumer, as a recommended learning model, a learning model among the stored learning models stored in the storage module that satisfies predetermined recommendation conditions, wherein the predetermined recommendation conditions may be conditions that are similar to the target learning model at a first similarity or higher.

[0018] In addition, the recommendation module calculates the recommended learning model according to a predetermined calculation method if a predetermined first additional condition is met even if the predetermined recommendation condition is not met, and the predetermined first additional condition may be a condition that the target learning model is similar to the target learning model at a second similarity level or higher, is similar to the target learning model at a level lower than the first similarity level, and the number of clusters of the stored learning models existing at a level lower than the second similarity level based on the target learning model is less than a predetermined number.

[0019] In addition, the predetermined calculation method may be a method in which, when there is one cluster of stored learning models that exist with the second similarity or higher based on the target learning model, the stored learning model that is closest to the average of the cluster is selected as the recommended learning model.

[0020] In addition, the predetermined calculation method may be a method of selecting the recommended learning model by a combination of the stored learning models that is closest to the average of each group when the number of groups of the stored learning models that exist with the second similarity or higher based on the target learning model is more than one and less than a predetermined number.

[0021] In addition, the recommendation module may select the recommended learning model based on the stored learning model for which a predetermined second additional condition is satisfied even if the predetermined recommendation condition and the predetermined first additional condition are not satisfied, and the predetermined second additional condition may be a condition that when the target learning model is connected to a stored learning model that is similar to the target learning model but less than the second similarity, the similarity with the target learning model is equal to or greater than the first similarity.

[0022] In addition, when any learning models are connected to each other, the system further includes a simulation module that simulates the state of the final learning model, and when multiple learning models are input into the simulation model, the simulation module can generate a simulation model through deep learning that calculates the state of the final learning model.

[0023] The system may further include a learning suggestion module that suggests to the administrator to generate a learning model if predetermined learning conditions are met, and the predetermined learning conditions may be conditions for which the predetermined recommendation conditions are not met.

[0024] The system further includes a price calculation module that calculates the fee to be charged while recommending the stored learning model, and the price calculation module can calculate the fee to be charged based on the similarity between the target learning model and the recommended learning model to be recommended.

[0025] A learning model recommendation method according to one embodiment of the present invention is realized by a learning model recommendation device and recommends a learning model calculated by machine learning or deep learning in response to a consumer's request, and includes a request receiving step in which a receiving module receives the consumer's request and calculation conditions of a target learning model, which is the learning model requested by the consumer; a condition storage step in which a storage module stores a stored learning model, which is a learning model previously generated, and the calculation conditions of the stored learning model; a similarity determination step in which a similarity determination module determines the similarity between the target learning model and the stored learning model; and a predetermined recommendation condition of the stored learning model stored in the storage module by the recommendation module may be a condition similar to the target learning model with a first similarity or higher.

[0026] In addition, the recommendation step may be a step of selecting or calculating a learning model to be recommended to the consumer by determining whether a predetermined first additional condition, which determines whether a condition is satisfied based on a second similarity lower than the first similarity even if the predetermined recommendation condition is not satisfied, and whether a predetermined second additional condition, which determines whether a condition is satisfied based on a virtual learning model generated by combining a plurality of the stored learning models, is satisfied. [Effects of the Invention]

[0027] The resource-aware monitoring device and the cloud integrated operation system including the same according to the present invention can reduce social resource costs.

[0028] In addition, the system can be operated efficiently.

[0029] Furthermore, development time can be significantly reduced.

[0030] Furthermore, system overload can be prevented.

[0031] However, the effects of the present invention are not limited to the effects described above, and any unmentioned effects can be clearly understood by a person having ordinary skill in the art to which the present invention pertains from this specification and the accompanying drawings. [Brief explanation of the drawings]

[0032] [Figure 1] 1 is a relationship diagram of a cloud integrated operation system according to an embodiment of the present invention. [Figure 2] 1 is a configuration diagram of a cloud integrated operation system according to an embodiment of the present invention. [Figure 3] 1 is a diagram illustrating the configuration relationship of a monitoring device of a cloud integrated operation system according to an embodiment of the present invention. [Figure 4] 1 is a flowchart of a learning model recommendation method implemented by a learning model recommendation device of a cloud integrated operation system according to one embodiment of the present invention. [Figure 5] 1 is a diagram illustrating predetermined recommendation conditions and a predetermined first additional condition in a learning model recommendation method according to one embodiment of the present invention. [Figure 6] FIG. 10 is a diagram illustrating a learning model mediation interface screen provided by an interface module of a learning model recommendation device according to one embodiment of the present invention. [Figure 7] 10 is a flowchart illustrating a process in which a learning model recommendation device according to an embodiment of the present invention selects a learning model to recommend to a monitoring device. [Figure 8] 5 is a diagram illustrating a method for calculating a first operating time and a second operating time by a model selection module of a monitoring device according to an embodiment of the present invention. FIG. DETAILED DESCRIPTION OF THE INVENTION

[0033] Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. However, the concept of the present invention is not limited to the embodiments shown, and a person skilled in the art who understands the concept of the present invention may easily propose other inventions or other embodiments within the concept of the present invention by adding, changing, or deleting other components within the same concept, which are also included within the concept of the present invention.

[0034] FIG. 1 is a diagram showing the relationship of a cloud integrated operation system according to one embodiment of the present invention. Referring to Figure 1, a cloud integrated operation system 10 according to one embodiment of the present invention may be a system that manages and operates virtual machines running on physical servers, and sells, rents, and provides cloud services to cloud users (demandors) who require learning models.

[0035] For this purpose, the cloud integrated operation system 10 is connected to the consumer M10 and the physical server S10 via a wired / wireless network, and can transmit and receive necessary information.

[0036] As an example, the consumer M10 may be a recipient of a virtualized cloud service.

[0037] As an example, a consumer M10 can send the conditions and request for a learning model to be purchased or rented to the learning model recommendation device, and the learning model recommendation device can provide the consumer with a learning model, an interface for requesting and sending / receiving a learning model, etc.

[0038] The cloud integrated operation system 10 may be a system that, when requested by a consumer (client), recommends a necessary cloud service or learning model in response to the request.

[0039] Furthermore, the learning model recommendation method can provide an interface so that the learning model can be used in a cloud environment.

[0040] The network in this invention may be a core network integrated with a wired public network, a wireless mobile communication network, or the mobile Internet, and may refer to a global open computer network structure that provides the TCP / IP protocol and multiple services existing in its upper layers, i.e., HTTP (Hyper Text Transfer Protocol), HTTPS (Hyper Text Transfer Protocol Secure), Telnet, FTP (File Transfer Protocol), TransDomain Protocol, etc., and is not limited to these examples, but comprehensively refers to a data communication network that can send and receive data in various forms.

[0041] The cloud integrated operation system is explained in detail below. FIG. 2 is a configuration diagram of a cloud integrated operation system according to one embodiment of the present invention, and FIG. 3 is a configuration relationship diagram of a monitoring device of the cloud integrated operation system according to one embodiment of the present invention.

[0042] Referring to Figures 2 and 3, a cloud integrated operation system according to one embodiment of the present invention may include a learning model recommendation device 100 that recommends a learning model in response to a consumer request, a management device that manages virtual machines running on a physical server, and a monitoring device that monitors the operation of the virtual machines.

[0043] Each device will be described in detail below. A learning model recommendation device 100 according to one embodiment of the present invention is a learning model recommendation device 100 that recommends a learning model calculated by machine learning or deep learning in response to a consumer's request, and includes a receiving module 110 that receives the consumer's request and calculation conditions of a target learning model, which is the learning model requested by the consumer, a first storage module 120 that stores a stored learning model, which is a learning model generated in the past, and the calculation conditions of the stored learning model, a similarity determination module 130 that determines the similarity between the target learning model and the stored learning model, and a recommendation module 140 that recommends to the consumer, from the stored learning models stored in the first storage module 120, a learning model that satisfies predetermined recommendation conditions as a recommended learning model.

[0044] In addition, the learning model recommendation device 100 may further include a simulation module 150 that simulates the state of a final learning model when any learning models are connected to each other.

[0045] In addition, the learning model recommendation device 100 may further include a learning suggestion module 160 that suggests to the administrator to generate a learning model if a predetermined learning condition is met.

[0046] In addition, the learning model recommendation device 100 may further include a price calculation module 170 that calculates a fee to be charged while recommending the stored learning model.

[0047] In addition, the learning model recommendation device 100 may further include an interface module 180 that calculates and transmits an interface that can display and collect information necessary for the learning model recommendation method on the computing device of an administrator or consumer.

[0048] The receiving module 110 can collect information necessary for the learning model recommendation method to be implemented.

[0049] For example, the receiving module 110 can receive from the consumer request information and calculation conditions, which are conditions for the learning model required by the consumer.

[0050] A consumer can input a learning model request and calculation conditions through an interface provided by the interface module 180 via his / her computing device.

[0051] The calculation conditions may include the learning purpose of the learning model, the type of calculation object, the learning method, the type of learning model, the amount of learning data, the type of learning data, and the learning time.

[0052] The learning objective can refer to the purpose or problem that the learning model is trying to solve.

[0053] For example, the learning purpose may be to identify a person based on video data, or to identify the presence or absence of lung cancer based on X-ray data.

[0054] However, the types of learning objectives are not limited to these, and can be varied in a variety of ways that are obvious to ordinary engineers.

[0055] The type of calculation may refer to the specific meaning of the result value.

[0056] For example, it can represent "person or no person", "lung cancer or no lung cancer", "prescribed treatment method", etc.

[0057] However, the types of calculations are not limited to these, and can be varied in a variety of ways that are obvious to ordinary engineers.

[0058] The learning method may refer to a machine learning and / or deep learning method for the learning model.

[0059] As an example, the learning method may be classified as deep learning or machine learning.

[0060] Furthermore, learning methods can be classified into supervised learning, unsupervised learning, and reinforcement learning.

[0061] The type of learning model can refer to the type of specific algorithm used for learning.

[0062] As an example, types of learning models can be classified into K-nearest neighbor algorithms (K-NN), support vector machine algorithms (SVM), decision tree algorithms, random forest algorithms, and the like.

[0063] As an example, types of learning models can be divided into ANN, CNN, DNN, GAN, and RNN.

[0064] However, the types of learning models are not limited to these, and can be varied in a variety of ways that are obvious to ordinary engineers.

[0065] The amount of training data can refer to the amount of data required to generate a training model.

[0066] As an example, the amount of training data can be expressed as a number and unit that represents the amount of data, such as GB or TB.

[0067] The type of training data can be divided into image format, text format, and video format.

[0068] Training time can refer to the time required for machine learning / deep learning.

[0069] However, the calculation conditions are not limited to these, and can be variously modified at a level that is obvious to an ordinary engineer.

[0070] The receiving module 110 may transmit the received information to other modules, including the similarity determination module 130 .

[0071] The receiving module 110 can receive information required to operate the learning model recommendation method and the learning model intermediation platform.

[0072] The first storage module 120 may store all information necessary to implement the learning model recommendation method.

[0073] The first storage module 120 may store a stored learning model, which is a learning model that has been learned in the past, in a matched state with the calculation conditions of the stored learning model.

[0074] The first storage module may include an internal memory and / or an external memory.

[0075] As an example, the built-in memory may include at least one of volatile memory (e.g., DRAM, SRAM, or SDRAM), non-volatile memory (e.g., OTPROM (one time programmable ROM, PROM, EPROM, EEPROM, mask ROM, flash ROM) flash memory, a hard drive, or a solid state drive (SSD)).

[0076] The external memory may include a flash drive (e.g., CF (Compact Flash), SD (Secure Digital), Micro-SD, Mini-SD, xD (Extreme Digital), MMC (Multimedia Card), or Memory Stick, etc.

[0077] The similarity determination module 130 may be a model that determines the similarity between the subject learning model and the storage learning model.

[0078] The similarity can be calculated by comparing the calculation conditions of the target learning model and the calculation conditions of the storage learning model.

[0079] The similarity determination module 130 can preprocess and vectorize the calculation conditions of the storage learning model. The vectorization method utilizes a known technique, and detailed description thereof may be omitted.

[0080] Similarly, the similarity determination module 130 can preprocess and vectorize the calculation conditions of the target learning model.

[0081] The similarity determination module 130 can calculate the similarity by comparing the calculation conditions of the target learning model with the calculation conditions of the target learning model.

[0082] For example, the similarity determination module 130 may calculate the similarity based on the distance between the calculation condition vector information of the target learning model and the calculation condition vector information of the storage learning model.

[0083] However, the method of determining the similarity of the similarity determination module 130 is not limited to this, and can be modified in various ways that are obvious to those skilled in the art.

[0084] The recommendation module 140 may select a recommended learning model by determining whether a predetermined recommendation condition, a predetermined first additional condition, and a predetermined second additional condition are satisfied.

[0085] The predetermined recommendation condition may be a condition that the target learning model is similar to the target learning model at a first similarity or higher.

[0086] If the distance to the target learning model is equal to or less than the first distance, it can be determined that the similarity is equal to or greater than the first similarity.

[0087] The predetermined first additional condition may be a condition that the population of the stored learning model that is similar to the target learning model at a second similarity level or higher, similar to the target learning model at a level less than the first similarity level, and exists based on the target learning model at a level less than the second similarity level is a predetermined number or less.

[0088] When the distance to the target learning model is greater than the first distance and less than the second distance, it may be determined that the similarity is greater than the second similarity and less than the first similarity.

[0089] Here, the second distance may refer to a distance that is longer than the first distance.

[0090] Here, the predetermined number of crowds may be three.

[0091] However, the predetermined number of the crowd is not limited to this, and can be varied in various ways at a level that is obvious to an ordinary engineer.

[0092] The predetermined second additional condition may be a condition that when the target learning model and the stored learning model that are similar but less than the second similarity are connected, the similarity with the target learning model is equal to or greater than the first similarity.

[0093] When the distance to the target learning model exceeds the second distance, it may be determined that the similarity is less than the second similarity.

[0094] The storage learning model itself is similar to the target learning model at a similarity level less than the second similarity, but when multiple storage learning models are combined, there may be cases where the storage learning model has a similarity level of the first similarity or higher to the target learning model.

[0095] As an example, it can be assumed that the target learning model is a model that detects whether there is an intruder based on video data and calculates a response plan according to the type of intrusion. The first stored learning model may be a learning model that determines whether there is an intruder based on video data, the second stored learning model may be a learning model that determines the form of intrusion based on video data, and the third stored learning model may be a model that calculates a response plan based on the type and form of intrusion. Each of the first stored learning model to the third stored learning model is similar to the target learning model at a similarity level less than a second level, but when the first stored learning model to the third stored learning model are combined, they may be similar to the target learning model at a similarity level greater than or equal to a first level.

[0096] This allows the learning model that meets the needs of the consumer to be recommended by making the most of the stored learning modules stored in the storage module 120.

[0097] The simulation module 150 can connect the stored learning modules stored in the storage module 120 to each other to generate a virtual learning model.

[0098] If the above-mentioned predetermined recommendation conditions and the first predetermined additional condition are not met, the simulation module 150 can receive the calculation conditions of the target learning model from the receiving module 110 and all information regarding the stored learning model from the storage module 120.

[0099] The simulation module 150 can combine the storage learning module based on data input to the object learning model and data calculated from the object learning model.

[0100] As a specific example, the simulation module 150 may select a storage learning model (starting storage learning model) in which input data input to the target learning model matches input data of an arbitrary storage learning model.

[0101] In addition, the simulation module 150 can select a storage learning model (final storage learning model) in which the output data output from the target learning model matches the output data of any storage learning model.

[0102] Here, matching can mean that the data type input and the purpose for which the data is analyzed match at least in part.

[0103] In addition, the simulation module 150 can select a storage learning model (intermediate storage learning model) that uses data that matches the output data of the initial storage learning model as input data and data that matches the input data of the final storage learning model as output data of the storage learning model.

[0104] The number of intermediate storage learning models may be zero, one, or multiple. If there are multiple intermediate storage learning models, the output data of the preceding intermediate storage learning model and the input data of the subsequent intermediate storage learning model may be aligned to match each other, and the input data of the most preceding intermediate storage learning model may be aligned to match the output data of the starting storage learning model, and the output data of the most subsequent intermediate storage learning model may be aligned to match the input data of the ending storage learning model, so that the simulation module 150 can select the intermediate storage learning model.

[0105] The simulation model 150 can generate a simulation model through machine learning / deep learning that calculates the state of the final learning model when multiple learning models are input.

[0106] The machine learning may be, but is not limited to, a back propagation algorithm, which is an algorithm that updates the weights of a neural network using labeled data in the output layer.

[0107] Furthermore, since the deep neural network and the back propagation algorithm are well known in the art, detailed explanations thereof may be omitted.

[0108] the starting storage learning model selected by the simulation module 150; The terminal storage learning model and / or the mediated storage learning model can be input into the simulation model to generate a virtual learning model.

[0109] In addition, the simulation module 150 can calculate the calculation conditions of the virtual learning model based on the stored learning model input to the simulation model.

[0110] As an example, the learning time can be calculated as the sum of the learning times of the stored learning models input into the simulation model.

[0111] As an example, the amount of training data may be calculated as the sum of the pool of training data input into the simulation model.

[0112] The simulation module 150 may not be able to calculate the calculation conditions of a virtual learning model that cannot be estimated as a stock learning model.

[0113] The similarity determination module 130 may determine the similarity based only on the calculation conditions calculated when determining the similarity between the virtual learning model and the target learning model.

[0114] The learning suggestion module 160 can suggest to the administrator to generate a learning model if predetermined learning conditions are met.

[0115] The predetermined learning condition may be a condition in which the predetermined recommendation condition is not met.

[0116] As a specific example, the predetermined learning condition may be a condition that there are a predetermined number or more similar target learning models that do not satisfy the predetermined recommendation condition, the predetermined first additional condition, and the predetermined second additional condition.

[0117] The learning suggestion module 160 can receive the calculation conditions of the target learning model (hereinafter referred to as the rejected learning model) for which the predetermined recommendation conditions, the predetermined first additional conditions, and the predetermined second additional conditions are not satisfied from the recommendation module 140, and can store them.

[0118] The learning suggestion module 160 can vectorize and cluster the calculation conditions of the rejection learning model.

[0119] The method for clustering data uses a known technique, so a detailed description thereof may be omitted.

[0120] When a clustered group consists of a predetermined number or more of rejection learning models, the learning restriction module can reconstruct the average vector value of the group and suggest to the administrator to generate a learning model under the calculated calculation conditions.

[0121] For this purpose, the interface module 180 may calculate an interface in which the calculation conditions proposed for learning are displayed to the administrator and the administrator is inquired as to whether or not to accept the conditions.

[0122] The price calculation module 170 can calculate the cost of using, selling, or licensing a stored learning model in a cloud environment.

[0123] The price calculation module 170 can determine the cost according to the difficulty of creating the stored learning model.

[0124] For example, the price calculation module 170 may set a higher cost for a product with a higher manufacturing difficulty. Conversely, the price calculation module 170 may set a lower cost for a product with a lower manufacturing difficulty.

[0125] The price calculation module 170 can calculate the fee to be charged based on the similarity between the target learning model and the recommended learning model to be recommended.

[0126] Here, the similarity can be calculated based on the distance between the two models.

[0127] This allows consumers to receive compensation for any disadvantages that arise from not being able to use the learning model they desire.

[0128] The interface module 180 can calculate an interface that can display or collect information required by the administrator or consumer while implementing the learning model recommendation method, and provide it to the administrator or consumer's computing device.

[0129] The interface module 180 can calculate an interface for a learning model intermediation platform where learning models can be purchased and sold, and provide the interface to buyers or sellers.

[0130] A detailed description of the learning model intermediation platform will be provided below.

[0131] The interface module 180 can transmit information required to operate the learning model recommendation method and the learning model intermediation platform.

[0132] The interface module 180 is required to learn from the administrator, The rejected calculation conditions can be communicated to the seller of the learning model.

[0133] The learning model seller list can consist of sellers who use the learning model intermediary platform.

[0134] This will revitalize the learning model platform and learning model sales market and meet the needs of consumers.

[0135] The learning model recommendation device can not only recommend learning models required within the system to other devices within the system, but also recommend appropriate learning models to cloud service users or individual Internet users.

[0136] The management device 200 according to one embodiment of the present invention may include an operation module 210 that operates virtual machines on a physical server, a repair module 220 that corrects errors when the virtual machine is operating abnormally, a migration module 230 that calculates the optimal placement of virtual machines on the physical server and migrates them, and an intermediation module 240 that mediates between the learning model recommendation device and the monitoring device 300.

[0137] The operation module 210 can perform operations such as creating and deleting virtual machines in response to user requests.

[0138] The repair module 220 can repair the virtual machine when an error occurs in the virtual machine by a predetermined repair method.

[0139] The repair module 220 may store a database of response methods according to errors. The repair module 220 may receive logs and metadata of a virtual machine that has an error from the monitoring device 300 and correct the virtual machine error according to the response methods stored in advance.

[0140] However, the repair method of the repair module 220 is not limited to this, and can be modified in various ways at a level that is obvious to a person of ordinary skill in the art.

[0141] In one example, the repair module 220 may request the model running module 330 to calculate an appropriate repair method according to the error. This allows a supervised learning model stored in the model running module 330 to be run to calculate an appropriate repair method according to the error, and the calculated repair method may be transmitted from the model running module 330 to the repair module 220.

[0142] The migration module 230 can migrate a virtual machine to place the virtual machine on a physical machine.

[0143] For example, the migration module 230 may request the model running module 330 to predict the workload of a virtual machine, and a supervised learning model stored in the model running module 330 may be run to calculate the workload predicted to be generated in the cloud service. The virtual machine workload predicted by the model running module 330 may be transmitted to the migration module 230, and the migration module 230 may calculate an optimized placement of virtual machines that minimizes a pre-specified objective function based on the predicted virtual machine workload, and thereby place the virtual machines on physical servers.

[0144] In contrast, detailed description may be omitted within the scope of known techniques.

[0145] The intermediary module 240 can act as an intermediary between the monitoring device 300 and the learning model recommendation device so that the monitoring device 300 can appropriately receive a required learning model recommendation from the learning model recommendation device.

[0146] The mediation module 240 cooperates with the monitoring device 300 to receive information required by the operation module 210 , repair module 220 , and migration module 230 from the monitoring device 300 .

[0147] However, without being limited thereto, the monitoring device 300 can directly communicate with the learning model recommendation device to exchange information necessary for recommending a learning model.

[0148] A monitoring device 300 according to an embodiment of the present invention monitors the operation of at least one virtual machine running on a physical server within allocated resources of a cloud server integrated management system. The monitoring device 300 may include a transceiver module 310 that collects operation information generated by the virtual machine, a monitoring module 320 that monitors whether the virtual machine is operating non-ideally based on the operation information collected by the transceiver module 310, a model operation module 330 that operates a supervised learning model to assist the monitoring function of the monitoring module 320 based on a request from the monitoring module 320, and a schedule module 340 that determines the operation time of the supervised learning model using a predetermined timing determination method so that the supervised learning model can be operated within the allocated resources.

[0149] The cloud server integrated management system may further include a model selection module 350 that selects the supervised learning model that satisfies predetermined operating conditions.

[0150] Furthermore, the cloud server integrated management system may further include a second storage module 360 ​​that stores all information required to implement the monitoring method.

[0151] The transmitting and receiving module 310 can collect operation information occurring in the virtual machine.

[0152] In addition, the information required to implement the monitoring method can be transmitted and received between the external server, the client transmission / reception module 310, and the internal configuration of the cloud server integrated management system.

[0153] As an example, the operational information may include all of the metric data and log data generated by the virtual machine.

[0154] The monitoring module 320 can monitor in real time whether the virtual machine is operating normally based on the operation information.

[0155] For this purpose, the monitoring module 320 may store error patterns of operational information that may be determined to be abnormal, and if a similar error pattern is detected, it may be determined that an error has occurred in the corresponding virtual machine.

[0156] The monitoring module 320 can transmit information about the virtual machine in which the error occurred to the management device 200.

[0157] The monitoring module 320 can calculate an interface that can display information about the load generated by the virtual machine, the load status of the physical server, a list of the physical servers in operation, and the like to the administrator of the cloud server integrated management system.

[0158] In addition, the monitoring module 320 can generate an interface that allows a consumer to check the usage status of the virtual machine, the incurred cloud usage costs, and the like.

[0159] Here, the interface may be represented via visual images such as graphs.

[0160] The monitoring function in the present invention may include functions to calculate, detect, monitor, and predict all information generated by cloud services, such as monitoring abnormal signs of virtual machines, virtual machine load prediction, load prediction of cloud integrated operation systems, the amount of power generated by physical servers, and detection of abnormal signs of physical machines.

[0161] The model running module 330 can run a monitoring learning model, which is a learning model that helps the management device 200 to operate or that helps the monitoring function of the monitoring module 320.

[0162] As an example, the supervised learning model may include a first supervised learning model that predicts the load that will be generated in the cloud server integrated management system within a predetermined future period.

[0163] As an example, the supervised learning model may include a second supervised learning model that predicts the workload that will be generated by virtual machines on a physical server within a predetermined future time period.

[0164] As an example, the predetermined future period may be 30 days.

[0165] However, the predetermined future period is not limited to this, and can be variously modified at a level that is obvious to an ordinary engineer.

[0166] For example, the monitoring learning model determines whether a virtual machine has an error based on its operational information. A third supervised learning model may be provided to calculate solutions to resolve the errors.

[0167] The model execution module 330 stores supervised learning models.

[0168] The model selection module 350 may select the supervised learning model that satisfies predetermined operating conditions, and transmit the selected supervised learning model to the model operation module 330 .

[0169] The predetermined operating condition may be a condition in which the device is operated such that a load lower than a predetermined threshold is applied.

[0170] The model selection module 350 can modify the predetermined threshold based on the load predicted by the first supervised learning model.

[0171] The model selection module 350 may receive information about used resources among all resources in the cloud integrated operation system for a predetermined period stored in the storage module in order to calculate a predetermined threshold.

[0172] Alternatively, the model selection module 350 may receive information about spare resources of the cloud integrated operation system for a predetermined future period through the first supervised learning model of the model operation module 330.

[0173] Here, the surplus resources may refer to a value obtained by subtracting the used resources from the total resources.

[0174] As an example, the predetermined threshold may mean 70% of the average spare resource for a future or predetermined period of time.

[0175] However, the present invention is not limited to this, and the exact value of the predetermined threshold value can be varied in various ways as would be obvious to one of ordinary skill in the art.

[0176] The model selection module 350 may request a learning model recommendation device to recommend the supervised learning model that satisfies the predetermined operating conditions.

[0177] When the model selection module 350 requests a required supervised learning model from the learning model recommendation device, the model selection module 350 can also transmit the calculation conditions for the supervised learning model.

[0178] The model selection module 350 receives a supervised learning model (recommended learning model) from the learning model recommendation device and transmits it to the model running module 330, which can store the supervised learning model.

[0179] The schedule module 340 may determine the operating time based on the load predicted by the first supervised learning model.

[0180] Here, the operating period can be a term that includes not only the start time but also the progress of time.

[0181] As an example, the predetermined timing determination method may be a method of determining the operation timing when a load is generated at a predetermined rate or less based on the resources allocated to the cloud integrated operation system.

[0182] As a specific example, the schedule module 340 may divide a predetermined future period into predetermined segment periods.

[0183] Here, the interval period can be in hours, days, or weeks.

[0184] Here, the schedule module 340 may calculate average usage resources predicted to be generated in the cloud integrated operation system for each partition period.

[0185] The schedule module 340 can determine, as an operating period, a period in which the used resource calculated for each section period is equal to or less than a predetermined ratio based on the total resource.

[0186] Here, the predetermined ratio (first ratio) may be 10%.

[0187] However, the predetermined ratio is not limited to this, and can be variously modified at a level that is obvious to those skilled in the art.

[0188] For example, the predetermined timing determination method may be a method of determining the operation timing as a timing at which the least load is generated within the predetermined future period.

[0189] As a specific example, the schedule module 340 may divide a predetermined future period into predetermined segment periods.

[0190] Here, the interval period can be minutes, hours, days, or weeks.

[0191] Here, the schedule module 340 may calculate an average usage resource generated in the cloud integrated operation system for each partition period.

[0192] The schedule module 340 may determine the section period in which the average resource usage calculated for each section period is the smallest as the operating time.

[0193] The schedule module 340 can effectively solve the problem that the operation rate of the cloud integrated operation system is hindered due to the operation of the operation learning model by determining the operation time when the load on the cloud integrated operation system is low.

[0194] The second storage module 360 ​​may store all information necessary to implement the monitoring method.

[0195] The second storage module 360 ​​may also store information on the overall resources of the past cloud integrated operation system, information on resources used in the past, and the like.

[0196] The second storage module may include an internal memory and / or an external memory.

[0197] As an example, the embedded memory may include at least one of volatile memory (e.g., DRAM, SRAM, or SDRAM), non-volatile memory (e.g., one time programmable ROM (OTPROM), PROM, EPROM, EEPROM, mask ROM, flash ROM, hard drive, or solid state drive (SSD)).

[0198] The external memory may include a flash drive, such as a compact flash (CF), secure digital (SD), Micro-SD, Mini-SD, extreme digital (xD), multi-media card (MMC), or memory stick.

[0199] The storage module included in the learning model recommendation device can be defined as the first storage module 120, and the storage module included in the monitoring device 300 can be defined as the second storage module 360.

[0200] The learning model recommendation method and monitoring method realized by each module will be described in detail below. FIG. 4 is a flowchart of a learning model recommendation method implemented by a learning model recommendation device in a cloud integrated operation system according to one embodiment of the present invention.

[0201] Hereinafter, detailed description may be omitted to the extent that it overlaps with the above content. Referring to Figure 4, a learning model recommendation method according to one embodiment of the present invention is realized by a learning model recommendation device and recommends a learning model calculated by machine learning or deep learning in response to a consumer's request. The method includes a request receiving step in which a receiving module receives the consumer's request and calculation conditions of a target learning model, which is the learning model requested by the consumer; a condition storage step in which a storage module stores a stored learning model, which is a learning model generated in the past, and the calculation conditions of the stored learning model; a similarity determination step in which a similarity determination module determines the similarity between the target learning model and the stored learning model; and a recommendation step in which a recommendation module recommends to the consumer a learning model that satisfies predetermined recommendation conditions from the stored learning models stored in the storage module.

[0202] Here, the recommendation step may be a step of determining whether a predetermined first additional condition, which determines whether a condition is satisfied based on a second similarity lower than the first similarity even if the predetermined recommendation condition is not satisfied, and whether a predetermined second additional condition, which determines whether a condition is satisfied based on a virtual learning model generated by combining a plurality of the stored learning models, is satisfied, and selecting or calculating a learning model to be recommended to the consumer.

[0203] The consumer can use a computing device to input the calculation conditions for the target learning model through the interface received from the interface module.

[0204] The collection module can receive the calculation conditions of the target learning model from the consumer's computing device and send the calculation conditions to the similarity determination module.

[0205] The similarity determination module can determine the similarity between the target learning model and the stored learning model stored in the storage module.

[0206] The similarity determination module can transmit the similarity determination result between the target learning model and the stored learning model to the recommendation module.

[0207] FIG. 5 is a diagram illustrating predetermined recommendation conditions and a predetermined first additional condition in a learning model recommendation method according to an embodiment of the present invention.

[0208] Referring to FIG. 4 and FIG. 5(a), the recommendation module can determine whether there is a stored learning model that satisfies a predetermined recommendation condition.

[0209] The predetermined recommendation condition may be a condition that the target learning model is similar to the target learning model at a first similarity or higher.

[0210] Being similar to the first similarity or higher can mean being at a first distance X10 or less from the target learning model A10.

[0211] If there are multiple stored learning models that satisfy the predetermined recommendation conditions, the recommendation module can select the stored learning model that is most similar to the target learning model as the recommended learning model C10.

[0212] Even if there is no stored learning model that satisfies the predetermined recommendation condition, the recommendation module can calculate the recommendation learning model according to a predetermined calculation method if a predetermined first additional condition is met.

[0213] The predetermined first additional condition may be a condition that the population of the stored learning model that is similar to the target learning model at a second similarity level or higher, similar to the target learning model at a level less than the first similarity level, and exists based on the target learning model at a level less than the second similarity level is a predetermined number or less.

[0214] This may be to complement and correct when the calculation conditions for the target learning model are not accurately specified due to the consumer's inexperience.

[0215] Referring to FIG. 5(b) and FIG. 5(c), a first additional condition predetermined as one and three populations of the pooled learning model may be satisfied.

[0216] Here, a crowd can be defined as being formed when the number of stored learning models within a predetermined distance from the center point of the crowd is equal to or greater than a predetermined number.

[0217] Here, the predetermined distance and the predetermined number can be variously modified at a level that is obvious to those skilled in the art.

[0218] The predetermined calculation method may be a method in which, when there is one cluster of stored learning models that exist with the second similarity or higher based on the target learning model, the stored learning model that is closest to the average of the cluster is selected as the recommended learning model.

[0219] Similarity with a second similarity or higher can mean that the target learning model is located at a second distance X20 or less from the target learning model.

[0220] Referring to FIG. 5(b), a cluster may be formed within a distance greater than the first distance X10 and equal to or less than the second distance X20.

[0221] The recommendation module can calculate the calculation conditions by inversely calculating the vector of the mean point M10 of a group within the first distance X10 and the second distance X20 based on the target learning model A10, and select a recommended learning model.

[0222] The predetermined calculation method may be a method of selecting the recommended learning model by a combination of the stored learning models that is closest to the average of each group when there are more than one but not more than a predetermined number of groups of the stored learning models that exist with the second similarity or higher based on the target learning model.

[0223] Referring to FIG. 5(c), three clusters may be formed beyond the first distance X10 and within the second distance X20 based on the object learning model A10.

[0224] As an example, the calculation conditions can be calculated by weighting the average point (M11) of the calculation conditions of the storage learning model that constitutes the first group, the average point (M12) of the calculation conditions of the storage learning model that constitutes the second group, and the average point (M13) of the calculation conditions of the storage learning model that constitutes the third group according to distance.

[0225] Here, the closer the distance to the target learning model, the higher the weight may be.

[0226] As an example, assuming that the distance between the calculation condition average point (M11) of the storage learning model constituting the first crowd and the target learning model is "10", the distance between the calculation condition average point (M12) of the storage learning model constituting the second crowd and the target learning model is "20", and the distance between the calculation condition average point M13 of the storage learning model constituting the third crowd and the target learning model is "30", the weights of the first crowd, the second crowd, and the third crowd are set to 3:2:1, and the above-mentioned calculation conditions can be calculated.

[0227] In this way, the recommendation module can calculate the calculation conditions by performing an inverse vector calculation on the calculated average value, and select the calculation conditions as the recommended learning model.

[0228] FIG. 5 shows an example of vectorization on two axes, but the present invention is not limited to this. The calculation conditions can be vectorized on three axes to determine the similarity and whether the above-mentioned conditions are satisfied.

[0229] The recommendation module may select the recommended learning model based on the stored learning model that satisfies a predetermined second additional condition even if the predetermined recommendation condition and the predetermined first additional condition are not satisfied.

[0230] As a specific example, the recommendation module may select a virtual learning model that satisfies a predetermined second additional condition based on the virtual learning model transmitted from the simulation module as a recommended learning model.

[0231] If there is no stored learning model that satisfies the predetermined recommendation conditions, the first predetermined condition, and the second predetermined additional condition, the learning suggestion module can store the calculation conditions of the target learning model.

[0232] The interface model may also calculate and transmit to the consumer computing device an interface that displays a message to the consumer that there is no stored learning model similar to the target learning model.

[0233] FIG. 6 is a diagram showing a learning model mediation interface screen provided by the interface module of the learning model recommendation device according to one embodiment of the present invention.

[0234] Referring to FIG. 6, the interface module can calculate an interface for a learning model intermediation platform that mediates between sellers who generate and sell learning models and buyers who purchase learning models.

[0235] As an example, the specifications (calculation conditions) of a learning model created by a seller may be listed, and a price for the model may be established and displayed on the interface.

[0236] Here, consumers can select, purchase, rent and use the learning models they need for their programs just like shopping on the Internet.

[0237] Figure 7 is a flowchart illustrating a process by which a learning model recommendation device according to one embodiment of the present invention selects a learning model to recommend to a monitoring device, and Figure 8 is a diagram illustrating a method by which a model selection module of a monitoring device according to one embodiment of the present invention calculates a first operating time and a second operating time.

[0238] Referring to FIGS. 7 and 8, the model selection module may transmit the first operating time and the second operating time together when transmitting the calculation conditions to the learning model recommendation device.

[0239] The model selection module can calculate the first operating time and the second operating time while determining the operating time.

[0240] The first operating time may refer to the total time of a continuous partition period during which the average usage resource of the cloud integrated operating system is generated at or below the first rate.

[0241] The second operating time may mean a time during which the average resource usage of the cloud integrated operation system is generated at or below the second ratio, and may be a time that is a combination of the first operating time and a total consecutive time adjacent to the first operating time.

[0242] Referring to FIG. 8(a), as an example, if the partition period is in hours, the periods during which the average resource usage of the cloud integrated operation system is predicted to be calculated at or below the first ratio W10 may be the operating periods T12, T13, and T14, and the first operating time may be 3 hours.

[0243] Here, since the average resource usage in the section periods T11 and T15 adjacent to the operating period is higher than the second ratio W20, the second operating period may not exist.

[0244] Referring to Figure 8(b), as an example, if the partition period is in hours, the period during which the resources of the cloud integrated operation system are predicted to be calculated at or below the first ratio W10 may be the operating period (T12, T13, T14), and the first operating time may be 3 hours.

[0245] Furthermore, since the average resource usage between the operating period and the adjacent section periods T11 and T15 is lower than the second ratio W20, the second operating time may be 5 hours, which is the sum of the first operating time and 2 hours.

[0246] The second operating time may be somewhat difficult for the cloud integrated operation system, but it can refer to the period during which the learning model can be operated within an acceptable range.

[0247] The learning model recommendation device can receive a request for recommending a learning model, calculation conditions, and first and second operating times directly from the monitoring device or from an intermediary module.

[0248] In the above content, the subject learning model can be understood as a supervised learning model for which recommendations are desired.

[0249] The learning model recommendation device can calculate whether there is a stored learning model similar to the supervised learning model for which recommendation is requested, as described above, and detailed explanations for this may be omitted to the extent that they overlap with the content described above.

[0250] However, unlike the recommendation method mentioned above, the system internally recommends learning models. In this case, the operating time can be further taken into consideration to determine whether a recommended learning model exists.

[0251] The simulation module can simulate the time required when the stored learning model stored in the first storage module is run.

[0252] Here, the amount of data expected to be input can be received together with the calculation conditions received from the monitoring device.

[0253] The simulation module can communicate the predicted run time of the stored learning model to the recommendation module.

[0254] The recommendation module may calculate whether there is a stored learning model in which the predetermined recommendation condition, the predetermined first additional condition, or the predetermined second additional condition is satisfied in the stored learning module in which the first operating time is satisfied. A detailed description of this may be omitted due to the limitation of overlap with the above content.

[0255] Here, the expected operating time of the storage learning module that satisfies the predetermined second additional condition can be calculated by adding together the expected operating times of all the storage learning models used (for example, the start storage learning model and the end storage learning model).

[0256] If there is no stored learning model that satisfies the predetermined recommendation condition, the predetermined first additional condition, or the predetermined second additional condition in the stored learning module that satisfies the first operating time, the recommendation module may calculate whether there is a stored learning model that satisfies the predetermined recommendation condition, the predetermined first additional condition, or the predetermined second additional condition in the stored learning module that satisfies the second operating time. A detailed description of this may be omitted due to the limitation of overlap with the above content.

[0257] Here, the expected operating time of the storage learning module that satisfies the predetermined second additional condition can be calculated by adding together the expected operating times of all the storage learning models used (for example, the start storage learning model and the end storage learning model).

[0258] If there is no stored learning model that satisfies the predetermined recommendation conditions, the predetermined first additional conditions, or the predetermined second additional conditions within the first operating time and the second operating time, the absence of a recommended learning model can be communicated to the model selection module, and the calculation conditions of the supervised learning model can be stored in the learning suggestion module and used for the above-mentioned learning.

[0259] The model selection module may select the recommended learning model received from the learning model recommendation device as the supervised learning model and may transmit the selected model to the model operation module.

[0260] A monitoring method according to an embodiment of the present invention is a monitoring method for monitoring the operation of at least one virtual machine running on a physical server within allocated resources of a cloud integrated operation system via a monitoring device, the monitoring method comprising: a step of collecting operation information generated by the virtual machine by a transmitting / receiving module; a step of monitoring whether the virtual machine is operating non-ideally based on the operation information by a monitoring module; and a step of operating a monitoring learning model that assists the monitoring function of the monitoring module based on a request from the monitoring module by a model operating module. and a step of determining an operation time of the supervised learning model through a predetermined time determination method so that the supervised learning model can be operated within the allocated resources by a scheduling module.

[0261] In addition, the monitoring method may further include a step of determining the operation time based on the load (usage resources) predicted by a first monitoring learning model that predicts the load that will occur in the cloud integrated operation system during a predetermined future period by the schedule module.

[0262] The monitoring method may further include selecting, by a model selection module, a supervised learning model that satisfies a predetermined operating condition.

[0263] The monitoring method may further include changing the predetermined threshold based on a load predicted by the first supervised learning model by a model selection module.

[0264] In the accompanying drawings, in order to more clearly express the technical idea of ​​the present invention, components that are not relevant to or irrelevant to the technical idea of ​​the present invention are simply depicted or omitted. In the above, the configuration and features of the present invention have been described based on the embodiment of the present invention, but the present invention is not limited thereto, and it will be apparent to those skilled in the art to which the present invention pertains that various modifications or variations can be made within the spirit and scope of the present invention, and therefore such modifications or variations are within the scope of the appended claims. [Explanation of symbols]

[0265] 10: Cloud integrated operation system 100: Learning model recommendation device 200: Management device 300: Monitoring equipment

Claims

1. A monitoring device that monitors the operation of at least one virtual machine running on a physical server within allocated resources of a cloud integrated operation system, a transmission / reception module that collects operation information generated by the virtual machine; a monitoring module that monitors whether the virtual machine is operating non-ideally based on the operation information collected by the transmitting / receiving module; a model operation module that operates a supervision learning model that assists the monitoring function of the monitoring module based on a request from the monitoring module; and a schedule module that determines when the supervised learning model is to be run through a predetermined timing determination method so that the supervised learning model can be run within the allocated resources. A monitoring device characterized by

2. The supervised learning model is A first supervised learning model that predicts a load generated in the cloud integrated operation system within a predetermined future period; The schedule module: determining the operation time based on the load predicted by the first monitoring learning model; The monitoring device according to claim 1 .

3. The predetermined time point determination method includes: This is a method for determining the operation period when the load generated is below a predetermined rate based on the allocated resources. The monitoring device according to claim 2 .

4. The predetermined time point determination method includes: The method is to determine the operation time at a time when the least load will be generated within the predetermined future period. The monitoring device according to claim 3 .

5. a model selection module that selects the supervised learning model for which a predetermined operating condition is satisfied; The predetermined operating conditions are: This is a condition under which the device is operated so that a load lower than a predetermined threshold is applied. The monitoring system according to claim 1 .

6. The learning model is a first supervised learning model that predicts a load that will be generated in the cloud integrated operation system within a predetermined future period; The model selection module: Varying the predetermined threshold based on the load predicted by the first supervised learning model. The monitoring device according to claim 5 .

7. a management device that manages virtual machines running on the physical server; a monitoring device that monitors the operation of the virtual machine; a learning model recommendation device that recommends a supervised learning model necessary for the monitoring device to realize a monitoring function, The monitoring device includes: The system further includes a transmission / reception module that collects information generated by the virtual machine; a monitoring module that monitors whether the virtual machine operates non-ideally based on the information about the virtual machine collected by the transmission / reception module; a model operation module that operates the supervised learning model to assist the monitoring function of the monitoring module based on a request from the monitoring module; a schedule module that determines the operation time of the supervised learning model through a predetermined time determination method so that the supervised learning model can be operated within allocated resources; and a model selection module that selects the supervised learning model to be operated from the model operation module when predetermined operation conditions are met, The model selection module: requesting the learning model recommendation device to recommend the supervised learning model for which the predetermined operating condition is satisfied; The learning model recommendation device, A learning model that satisfies a predetermined recommendation condition, which is a condition similar to the supervised learning model by a first similarity or more, is inferred from among previously stored stored learning models. A cloud integrated operation system characterized by:

8. The learning model recommendation device, Even if there is no learning model that satisfies the predetermined recommendation condition among the stored learning models stored in advance, a learning model that satisfies a predetermined first additional condition that determines whether the condition is satisfied based on a second similarity that is lower than the first similarity, or a predetermined second additional condition that determines whether the condition is satisfied based on a virtual learning model generated by combining a plurality of the stored learning models, is recommended. The cloud integrated operation system according to claim 7.

9. The predetermined first additional condition is: A condition in which the supervised learning model is similar to the supervised learning model at a second similarity or more, is similar to the supervised learning model at a similarity less than the first similarity, and a group of the stored learning models that are stored in advance and exist at a similarity less than the second similarity based on the supervised learning model exists in a predetermined number or less. The cloud integrated operation system according to claim 8.

10. A monitoring method for monitoring, via a monitoring device, the operation of at least one virtual machine running on a physical server within allocated resources of a cloud integrated operation system, comprising: a step of collecting operation information generated in the virtual machine by a transmitting / receiving module; a step of monitoring whether the virtual machine is operating non-ideally based on the operation information by a monitoring module; The method includes a step of operating a supervised learning model that assists the monitoring function of the monitoring module based on a request from the monitoring module by a model operating module, and a step of determining a timing of the supervised learning model according to a predetermined timing determination method by a schedule module so that the supervised learning model can be operated within the allocated resources. A monitoring method characterized by:

Images