KR20230161122A - Provisioning method for cloud service and system thereof - Google Patents

Abstract

A provisioning method and system for cloud services are provided. A provisioning method for a cloud service according to some embodiments of the present disclosure monitors utilization information including the overall utilization of cloud resources, receives a request for a regular instance from a client, and determines the overall utilization rate. In response to the determination that the threshold value is greater than 1, at least some of the already provisioned spot instances may be terminated and the requested regular instances may be provisioned. By doing so, the efficiency of using cloud resources can be improved and the quality of cloud services can be guaranteed.

Description

Provisioning method and system for cloud services {PROVISIONING METHOD FOR CLOUD SERVICE AND SYSTEM THEREOF}

This disclosure relates to a provisioning method and system for cloud services, and more specifically, to a method for dynamically provisioning cloud resources and a system for performing the method to improve the use efficiency of cloud resources.

In cloud services, the provisioning process refers to a series of processes that allocate limited resources of physical cloud nodes according to clients' requests, create cloud instances, and provide the created cloud instances to the clients who requested them. do. Because this provisioning process determines the quality of cloud services and the efficiency of using cloud resources, many cloud service providers are still trying to devise efficient provisioning algorithms.

Meanwhile, most cloud service provision systems provide services using only regular instances whose usage period and performance are guaranteed. In addition, most cloud service provision systems do not consider the actual usage rate of regular instances and perform provisioning only by considering the number of provisioned instances (e.g. a method of setting the maximum number of instances that can be provisioned and performing provisioning only within that number). ). In this case, even though the actual usage rate of cloud resources is not high, additional regular instances cannot be provisioned, so the efficiency of using cloud resources and the profitability of the cloud service provider are bound to decrease.

Korean Patent Publication No. 10-2019-0059661 (published on May 31, 2019)

The technical problem to be solved through some embodiments of the present disclosure is to provide a provisioning method that can improve the efficiency of using cloud resources while ensuring the quality of cloud services and a system that performs the method.

Another technical problem to be solved through some embodiments of the present disclosure is a method for efficiently performing provisioning considering the characteristics of regular instances and spot instances, and a system for performing the method. is to provide.

The technical problems of the present disclosure are not limited to the technical problems mentioned above, and other technical problems not mentioned can be clearly understood by those skilled in the art from the description below.

A provisioning method for a cloud service according to some embodiments of the present disclosure to solve the above-described technical problem is a method performed on at least one computing device, and includes monitoring utilization information including the overall utilization of cloud resources. , receiving a request for a regular instance from a client and in response to determining that the overall utilization is greater than or equal to a first threshold, terminating at least some of the previously provisioned spot instances and using the requested regular instance. It may include provisioning an instance.

In one embodiment, the cloud resources include processor resources and memory resources, the first threshold includes a first resource threshold for the processor resource and a second resource threshold for the memory resource, and the requested regular Provisioning the instance may include terminating at least some of the instances in response to determining that the overall utilization of the processor resources is greater than or equal to the first resource threshold or the overall utilization of the memory resources is greater than or equal to the second resource threshold. You can.

In one embodiment, the overall utilization rate of cloud resources is the first overall utilization rate, and provisioning the requested regular instance includes specifying a termination target in the list of the already provisioned spot instances, terminating the specified spot instance. It may include predicting a second overall utilization of the cloud resources according to , and terminating the designated spot instance in response to determining that the second overall utilization is less than the first threshold.

In one embodiment, receiving a request for a spot instance from a client and, in response to determining that the overall utilization is below a second threshold, provisioning the requested spot instance, The second threshold may be greater than the first threshold.

In one embodiment, receiving requests for a plurality of Spot Instances from a client, wherein the requests are received within a current time window, and at the end of the current time window, provisioning the plurality of Spot Instances. It may further include steps of performing.

A provisioning system for a cloud service according to some embodiments of the present disclosure for solving the above-mentioned technical problem includes one or more processors and a memory that stores one or more instructions, wherein the one or more processors store the one or more instructions. Monitoring utilization information including the overall utilization of cloud resources by executing an instruction, receiving a request for a regular instance from a client, and in response to determining that the overall utilization is greater than or equal to a first threshold, An operation may be performed to terminate at least some of the already provisioned spot instances and provision the requested regular instances.

A computer program according to some embodiments of the present disclosure for solving the above-described technical problem includes the steps of combining with a computing device and monitoring utilization information including the overall utilization of cloud resources, receiving a regular instance from a client. Receiving a request for and in response to determining that the overall utilization is greater than or equal to a first threshold, terminating at least some of the already provisioned spot instances and provisioning the requested regular instances. It may be stored in a computer-readable recording medium for execution.

According to some embodiments of the present disclosure, provisioning may be performed based on the overall utilization of cloud resources. In this case, the occurrence of idle resources can be minimized and the efficiency of using cloud resources can be greatly improved.

In addition, by performing provisioning by reflecting the characteristics of regular instances and spot instances, the efficiency of using cloud resources can be improved and the quality of cloud services can be guaranteed. For example, when the overall utilization of cloud resources exceeds the first threshold, the quality of regular instances can be guaranteed by terminating spot instances and provisioning regular instances. Additionally, when the overall usage rate of cloud resources is less than the second threshold (however, the second threshold is usually greater than the first threshold), the usage efficiency of cloud resources can be improved by allowing provisioning of spot instances. Furthermore, when the overall utilization of cloud resources is higher than the second threshold, provisioning of spot instances is prohibited, so that the impact of spot instances on the performance of regular instances can be minimized.

Additionally, by provisioning a plurality of spot instances using a price bidding method, the profitability of the cloud service provider can be greatly improved.

Additionally, by monitoring the utilization of cloud nodes and cloud instances through the node agent and instance agent, utilization monitoring of cloud resources can be accurately performed.

In addition, by predicting the overall utilization of cloud resources following termination of specified spot instances and terminating spot instances only until the predicted overall utilization falls below the first threshold, excessive termination of already provisioned spot instances can be prevented. there is.

Additionally, termination processing for Spot Instances can be performed in a simple but fair manner by determining which Spot Instances will be terminated first based on criteria such as utilization, bid price, and/or usage period.

The effects according to the technical idea of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description below.

1 is an exemplary diagram for explaining a provisioning system and a cloud service providing system according to some embodiments of the present disclosure.
Figure 2 is an exemplary diagram for explaining a cloud resource monitoring method according to an embodiment of the present disclosure.
FIG. 3 is an exemplary diagram illustrating the relationship between a virtual cloud resource and a cloud instance that may be referenced in various embodiments of the present disclosure.
FIG. 4 is an exemplary diagram illustrating the timing and method of provisioning a regular instance according to an embodiment of the present disclosure.
FIG. 5 is an exemplary diagram illustrating the provisioning method and timing of spot instances according to an embodiment of the present disclosure.
FIG. 6 is an exemplary diagram for explaining the configuration of cloud nodes according to an embodiment of the present disclosure.
7 is an example flowchart illustrating a provisioning method for a cloud service according to some embodiments of the present disclosure.
Figure 8 is an example flowchart showing a regular instance provisioning method according to an embodiment of the present disclosure.
9 and 10 are exemplary diagrams for explaining a spot instance termination method according to an embodiment of the present disclosure.
Figure 11 is an example flowchart showing a spot instance provisioning method according to an embodiment of the present disclosure.
FIG. 12 is an exemplary diagram for further explaining a first threshold and a second threshold used in provisioning methods according to some embodiments of the present disclosure.
13 illustrates an example computing device that can implement a provisioning system in accordance with some embodiments of the present disclosure.

Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the attached drawings. The advantages and features of the present disclosure and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the technical idea of the present disclosure is not limited to the following embodiments and may be implemented in various different forms. The following examples are merely intended to complete the technical idea of the present disclosure and to cover the technical field to which the present disclosure belongs. is provided to fully inform those skilled in the art of the scope of the present disclosure, and the technical idea of the present disclosure is only defined by the scope of the claims.

In describing various embodiments of the present disclosure, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present disclosure, the detailed description will be omitted.

Unless otherwise defined, terms (including technical and scientific terms) used in the following embodiments may be used in a meaning that can be commonly understood by those skilled in the art in the technical field to which this disclosure pertains. It may vary depending on the intentions or precedents of engineers working in related fields, the emergence of new technologies, etc. The terminology used in this disclosure is for describing embodiments and is not intended to limit the scope of this disclosure.

The singular expressions used in the following embodiments include plural concepts, unless the context clearly specifies singularity. Additionally, plural expressions include singular concepts, unless the context clearly specifies plurality.

In addition, terms such as first, second, A, B, (a), (b) used in the following embodiments are only used to distinguish one component from another component, and the terms The nature, sequence, or order of the relevant components are not limited.

Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the attached drawings.

FIG. 1 is an exemplary diagram illustrating a provisioning system 10 and a cloud service providing system 11 according to some embodiments of the present disclosure.

As shown in FIG. 1, the cloud service providing system 11 according to embodiments may be a system that provides a cloud service to at least one client 13-1 to 13-n. For example, the cloud service provision system 11 provides a cloud service to the clients 13-1 to 13-n by provisioning a cloud instance in response to a request from the clients 13-1 to 13-n. (e.g. a service that provides computing resources) can be provided. Hereinafter, for convenience of explanation, the reference number '13' will be used when referring to an arbitrary client (e.g. 13-1) or to collectively refer to all clients (13-1 to 13-n).

As shown, the cloud service providing system 11 may be configured to include a provisioning system 10 and a plurality of cloud nodes 12-1 to 12-k. Hereinafter, each component of the cloud service provision system 11 will be described. However, for convenience of explanation, the reference number '12' is used when referring to an arbitrary cloud node (e.g. 12-1) or to collectively refer to all cloud nodes (12-1 to 12-k).

The provisioning system 10 may monitor cloud resources provided by cloud nodes 12 and provision cloud instances according to requests from clients 13. At this time, the provisioning system 10 may perform provisioning by considering the monitoring results. For reference, in this technical field, cloud instance may be used interchangeably with terms such as 'virtual instance', 'virtual machine', and 'virtual node'. However, since one cloud node 12 may be set to provide only one cloud instance, in some cases, the cloud instance may even encompass the concept of a 'physical machine'.

Cloud resources may be, for example, processor resources, memory resources, storage resources, network resources, etc., but the scope of the present disclosure is not limited thereto. In addition, monitoring information about cloud resources may include, for example, the overall utilization rate of cloud resources (e.g. the utilization rate of cloud resources provided by all cloud nodes 12), the resource utilization rate of the cloud node 12 (e.g. individual cloud node 12-1 resource utilization rate), or resource utilization rate of a cloud instance (e.g. resource utilization rate of an individual cloud instance), but the scope of the present disclosure is not limited thereto.

Provisioning system 10 may monitor cloud resources periodically or aperiodically. However, the specific monitoring method may vary depending on the embodiment.

In one embodiment, as shown in FIG. 2, the provisioning system 10 connects the cloud node 12 and each cloud instance 21-1 to 21-n through the instance agent 23 and the node agent 22. You can monitor (or collect) resource utilization. Here, the instance agent 23 is a module that operates on a cloud instance (e.g. 21-1), and measures and measures the resource utilization (e.g. virtual processor utilization, virtual memory utilization) of the cloud instance (e.g. 21-1). can provide results. Of course, the instance agent 23 may further perform measurements on other monitoring metrics. Additionally, the node agent 32 is a module that operates on the cloud node 12 and can measure the resource utilization rate of the cloud node 12 and provide the measured results. According to this embodiment, the resource utilization of the cloud node 12 and the cloud instance (e.g. 21-1) can be monitored very accurately through the different types of agents 22 and 23.

Additionally, the provisioning system 10 may allocate cloud resources according to the request of the client 13 and create a cloud instance. For example, as shown in FIG. 3, the provisioning system 10 allocates virtual processors 32-1 to 32-k and virtual memories 31-1 to 31-n provided by the cloud node 12. Thus, a cloud instance 33 can be created and the created cloud instance 33 can be provided to the client 13. At this time, the maximum number of virtual processors and virtual memory that the cloud node 12 can provide (or that can be allocated to the cloud node 12) may be preset fixed values or may be values that vary depending on the situation. FIG. 3 shows an example where the maximum number of virtual processors that the cloud node 12 can provide is 'K' and the maximum number of virtual memories is 'N'. In addition, Figure 3 shows as an example that a cloud instance (e.g. 33) is created using one virtual processor (e.g. 32-1) and one virtual memory (e.g. 31-1). However, the scope of the present disclosure is not limited to these examples. For example, the provisioning system 10 may create a cloud instance (e.g. 33) by allocating a plurality of virtual processors (e.g. 32-1) and a plurality of virtual memories (e.g. 31-1). In the present disclosure, the virtual processor (e.g. 32-1) may be referred to as a 'virtual processor resource', and the virtual memory (e.g. 31-1) may be referred to as a 'virtual memory resource'. Additionally, virtual processors (e.g. 32-1) and virtual memory (e.g. 31-1) may be referred to as 'virtual cloud resources'.

In the previous example, the maximum number of virtual processors and virtual memory may be set to different values or may be set to the same value. Additionally, the maximum number of virtual processors and virtual memories may be set to the same value for each cloud node 12, or may be set to different values. In addition, the amount (size) of resources possessed by each virtual processor or each virtual memory may be the same or different.

Meanwhile, in various embodiments of the present disclosure, the provisioning system 10 may perform provisioning for two types of cloud instances. In other words, the cloud service provision system 11 can provide cloud services using two types of cloud instances. Specifically, the provisioning system 10 can perform provisioning for regular instances and spot instances by reflecting the characteristics of each instance. The two instances are cloud instances that provide different Service Level Agreements (SLAs). Regular instances refer to instances where the requested usage period and performance are guaranteed, and spot instances refer to instances where the requested usage period and performance are not guaranteed (i.e. , may refer to an instance that may be terminated unintentionally. If you are working in the relevant technology field, you will already be familiar with the concepts of regular instances and spot instances, so description of them will be omitted.

In the previous embodiments, since regular instances and spot instances are instances with different characteristics, the provisioning time, method, and/or algorithm of the two instances may be designed differently.

For example, as shown in FIG. 4, the provisioning system 10 can perform immediate provisioning for regular instances. In other words, when a request 41 for a regular instance is received from the client 13, the provisioning system 10 may immediately provision the requested regular instance and provide a performance result 42.

Additionally, for example, as shown in FIG. 5, the provisioning system 10 may not perform immediate provisioning for spot instances. Specifically, provisioning system 10 may suspend processing (i.e., perform provisioning) on spot instance requests 51 and 52 received from clients 13 until the end point of the current time window is reached. (e.g. Insert requests 51 and 52 into the request queue and wait). And, at the last point of the current time window, the provisioning system 10 may perform (attempt) batch provisioning for the requested spot instances and provide a performance result 53. For example, the provisioning system 10 may perform provisioning based on monitoring results for cloud resources and the bid price of the requested spot instance. In this case, the effect of improving the profitability of cloud service providers can be achieved due to price competition for spot instances.

In the previous example, if the allowable waiting time for a particular request (e.g. 51) is exceeded before reaching the last point of the current time window, provisioning system 10 excludes that request (e.g. 51) and waits for the remaining requests (e.g. 52). Provisioning can only be performed for . Here, the allowable waiting time is a value set by the client 13 and may mean the maximum time that can wait until the requested spot instance is provisioned.

Additionally, the length of the time window may be a preset fixed value or a value that may vary depending on the situation. For example, the length of the time window may be a value that changes based on the frequency of occurrence of spot instance requests, the overall utilization rate of cloud resources, changes in the overall utilization rate, the average value of the allowable waiting time, etc. (e.g., if the overall utilization rate is consistently high, the length of the time window may be changed. Set the length to be longer, and vice versa to be shorter; the length of the time window can be adjusted taking into account the average value of the acceptable latency).

The specific method by which the provisioning system 10 provisions regular instances and spot instances will be described in detail with reference to the drawings of FIG. 7 and below.

Provisioning system 10 may be implemented with at least one computing device. For example, all functions of provisioning system 10 may be implemented in a single computing device, with a first function of provisioning system 10 implemented in a first computing device and a second function implemented in a second computing device. It could be. Alternatively, specific functions of provisioning system 10 may be implemented with a plurality of computing devices.

The computing device may include any device equipped with a computing function. Refer to FIG. 13 for an example of such a device.

Description will be made again with reference to FIG. 1 .

Cloud node 12 may refer to a physical device that provides cloud resources. For example, cloud node 12 may provide processor resources and memory resources. At this time, processor resources may be provided divided into multiple virtual processors (i.e., logically divided), and memory resources may also be provided divided into multiple virtual memories. And, as provisioning is performed, at least one cloud instance may be executed on the cloud node 12 (see FIG. 3). In some cases, the cloud node 12 may be referred to by terms such as 'cloud server', 'cloud device', 'computing node', 'resource node', 'resource device', or 'physical machine'.

The cloud node 12 may also be any device equipped with a computing function, and reference is made to FIG. 13 for an example of such a device.

Meanwhile, in one embodiment, as shown in FIG. 6, the cloud nodes 12 include computing nodes 12-1 to 12-n that provide cloud resources and these (12-1 to 12-n) It may be composed of a controller node 61 that controls (e.g. controls cloud instance creation, termination, etc.). In this case, the provisioning system 10 may perform provisioning by delivering the performance result of the provisioning algorithm to the controller node 61 to create the requested cloud instance.

So far, the configuration and operation of the provisioning system 10 and the cloud service providing system 11 have been described with reference to FIGS. 1 to 6. Hereinafter, various methods that can be performed in the provisioning system 10 will be described with reference to FIGS. 7 to 13.

Hereinafter, for convenience of understanding, the description will be continued assuming that the methods to be described later are performed in the provisioning system 10 in the environment illustrated in FIG. 1. Additionally, the description will be continued assuming that all steps/operations of the methods to be described later are performed by the provisioning system 10. Accordingly, when the subject of a specific step/action is omitted, it can be understood as being performed by the provisioning system 10. However, in a real environment (or an environment other than that illustrated in FIG. 1), some steps/operations of the method to be described later may be performed on another computing device.

7 is an example flowchart schematically illustrating a provisioning method for a cloud service according to some embodiments of the present disclosure. However, this is only a preferred embodiment for achieving the purpose of the present disclosure, and of course, some steps may be added or deleted as needed.

As shown in FIG. 7, the provisioning method according to embodiments may begin at step S71 of monitoring cloud resources. For example, the provisioning system 10 may measure the overall utilization of cloud resources (e.g. the utilization of the entire cloud resources provided by cloud nodes 12), the resource utilization of individual cloud nodes (e.g. 12) or individual cloud instances (e.g. 31-1), etc. can be monitored. Cloud resources may be, for example, processor resources, memory resources, etc. However, it is not limited to this.

The above-described resource utilization rate (e.g. overall utilization rate, resource utilization rate of cloud instance, etc.) may be a value measured at a specific point in time, or may be an average value of the utilization rate measured during a unit time (i.e., a certain time period). At this time, the unit time may be a preset fixed value or a value that changes depending on the situation. For example, when the utilization rate of cloud resources changes rapidly or the frequency of cloud instance requests increases rapidly, the provisioning system 10 may set the unit time to a smaller value for more accurate monitoring. In the opposite case, the unit time may be set to a larger value.

At step S72, a request for a cloud instance (or service) may be received. For example, the provisioning system 10 may receive a provisioning request for at least one cloud instance from at least one client 13.

A request for a cloud instance may include information on the type of cloud instance (i.e., information indicating a regular instance or a spot instance) as well as specification information on the requested service (or instance). And, this information can be used to perform provisioning of the corresponding cloud instance.

For example, if the requested cloud instance is a regular instance, the service specification information may include the usage period of the instance, required performance, images to be run on the instance, allowable price, amount of processor resources, amount of memory resources, etc. However, the scope of the present disclosure is not limited thereto.

Additionally, for example, if the requested cloud instance is a spot instance, the service specification information may include the instance's usage period, required performance, images to run on the instance, bid price, amount of processor resources, amount of memory resources, and allowable waiting time (i.e. , maximum time to wait until provisioning), etc., but the scope of the present disclosure is not limited thereto.

At step S73, the type of cloud instance requested may be determined. If the requested cloud instance is a regular instance, step S74 may be performed, and if the requested cloud instance is a spot instance, step S75 may be performed.

In step S74, provisioning for the requested regular instance may be performed according to the first provisioning algorithm. The detailed process of this step (i.e., the first provisioning algorithm) will be described in detail later with reference to FIGS. 8 to 10.

In step S75, provisioning for the requested spot instance may be performed according to the second provisioning algorithm. The detailed process of this step (i.e., the second provisioning algorithm) will be described in detail later with reference to FIGS. 11 and 12.

In step S76, provisioning performance results may be provided. For example, the provisioning system 10 may provide provisioning performance results (e.g. success, failure, information on provisioned cloud instances, etc.) to the client 13 that requested a cloud instance. The provisioning system 10 may store provisioning results in a database.

So far, a provisioning method for a cloud service according to some embodiments of the present disclosure has been schematically described with reference to FIG. 7 . Hereinafter, a regular instance provisioning method according to an embodiment of the present disclosure will be described with reference to FIGS. 8 to 10.

Figure 8 is an example flowchart showing a regular instance provisioning method according to an embodiment of the present disclosure. However, this is only a preferred embodiment for achieving the purpose of the present disclosure, and of course, some steps may be added or deleted as needed.

As shown in FIG. 8, the provisioning method according to the embodiment may begin at step S81 to determine whether cloud resources are sufficient. If it is determined that the cloud resources are sufficient, step S82 may be performed, and in the opposite case, step S83 may be performed.

For example, the provisioning system 10 may determine whether cloud resources are sufficient based on the number of cloud instances that can be created (provisioned) and/or the number of virtual cloud resources (e.g. virtual processors, virtual memory) that can be allocated. As a more specific example, the provisioning system 10 may determine that the cloud resources are sufficient when the number of virtual processors that can be allocated is greater than or equal to a predetermined threshold and the number of virtual memories is also greater than or equal to a predetermined threshold. Here, the number of virtual processors (or virtual memory) that can be allocated is the number of virtual processors (or virtual memory) already allocated from the maximum number of virtual processors (or virtual memory) that can be allocated through all cloud nodes 12. It can be calculated by doing. As described above, the maximum number of virtual processors (or virtual memories) that can be allocated may be a kind of set value (i.e., the value of a settable parameter). As another example, the provisioning system 10 may determine whether the cloud resources are sufficient by comparing the total usage rate of the cloud resources with a predetermined threshold.

In step S82, it may be determined whether the overall usage rate of cloud resources is greater than or equal to the first threshold TH1. In response to determining that the first threshold is above, step S84 may be performed, and vice versa, step S85 may be performed. Here, the first threshold is a value that determines whether to terminate a previously provisioned spot instance, and may be set for each type of cloud resource. And, the first threshold may be set to the same value or may be set to different values for different resource types.

When the first threshold is set for each type of cloud resource, step S84 may be performed if any one of the total usage rates of each cloud resource is greater than or equal to the first threshold. For example, assume that the first threshold consists of a first resource threshold for processor resources and a second resource threshold for memory resources. In this case, when the total utilization of processor resources is greater than or equal to the first resource threshold or the total usage of memory resources is greater than or equal to the second resource threshold, the provisioning system 10 may perform step S84. Hereinafter, for convenience of understanding, 'first threshold', 'first resource threshold', and 'second resource threshold' will continue to be used with the same meaning as described above.

This will be described again with reference to FIG. 8 .

At step S83, provisioning for the requested regular instance may be determined to have failed.

At step S84, the spot instance may be terminated. Specifically, at least one instance designated as a termination target among pre-provisioned spot instances may be terminated, and methods for specifying the termination target may be designed in various ways. The detailed process of this step will be described in detail with reference to FIGS. 9 and 10 later.

For reference, if it is impossible to terminate a Spot Instance (e.g., if there are no Spot Instances to terminate, if the overall utilization of cloud resources is greater than the first threshold even if all Spot Instances are terminated, etc.), provisioning will be determined to have failed. Can (see S83)

At step S85, the requested regular instance may be provisioned. For example, the provisioning system 10 may create a requested regular instance using a virtual processor and virtual memory provided by a specific cloud node 12.

At step S86, provisioning for the requested regular instance may be determined to be successful.

So far, the regular instance provisioning method according to an embodiment of the present disclosure has been described with reference to FIGS. 8 and 9. According to the above, by performing provisioning for regular instances based on the overall utilization of cloud resources, the occurrence of idle resources can be minimized and the use efficiency of cloud resources can be greatly improved. In addition, when the overall utilization of cloud resources exceeds the first threshold, the quality of regular instances can be guaranteed by terminating spot instances and provisioning regular instances.

Hereinafter, a spot instance termination method according to an embodiment of the present disclosure will be described with reference to FIGS. 9 and 10.

Figure 9 is an example flowchart showing a spot instance termination method according to an embodiment of the present disclosure. However, this is only a preferred embodiment for achieving the purpose of the present disclosure, and of course, some steps may be added or deleted as needed.

As shown in FIG. 9, the method for terminating a spot instance according to an embodiment may begin at step S91 of specifying a termination target from a list of pre-provisioned spot instances. For example, the provisioning system 10 may specify a termination target based on the resource utilization rate, bid price, usage period, etc. of individual spot instances.

As a more specific example, the provisioning system 10 may specify the spot instance with the highest resource utilization (e.g. utilization of processor resources or memory resources) as the termination target. Alternatively, the provisioning system 10 may designate the spot instance with the smallest difference between the total predicted usage rate according to termination (see description below) and the first threshold as the termination target.

As another example, provisioning system 10 may specify a termination target based on the bid price. For example, the spot instance with the lowest bid price may be designated for termination. In this case, price competition for spot instances can be induced, ultimately improving the profitability of cloud service providers.

As another example, the provisioning system 10 may specify a termination target based on the period of use. For example, Spot Instances with the oldest usage period may be designated as conventional targets.

As another example, the provisioning system 10 may randomly designate a termination target.

As another example, a termination target may be designated based on various combinations of the above-described examples. For example, the provisioning system 10 may designate the instance with the highest resource utilization among spot instances whose bid price is below the standard as the termination target.

In step S92, the overall utilization of cloud resources upon termination of the specified instance may be predicted. For example, the provisioning system 10 can predict the overall utilization of processor resources and memory resources according to instance termination. Specifically, the provisioning system 10 predicts a decrease in overall resource utilization due to termination of a specified instance, and reflects the predicted decrease in the overall utilization of current cloud resources, thereby calculating a predicted value of overall utilization due to termination of a specified instance. You can. For example, the provisioning system 10 can predict the overall utilization of processor resources according to termination of a specified instance according to Equation 3 below, and can also predict the overall utilization of memory resources due to termination of a designated instance in a similar manner.

In Equation 3 above, 'E_AVG_10m_pCPU' and 'AVG_10m_pCPU' mean the total utilization of processor resources predicted according to instance termination and the total utilization of current processor resources (e.g. assumed to be an average value for a unit time of 10 minutes), respectively. In addition, 'AVG_10m_vCPU' refers to the processor resource utilization rate of the specified spot instance, and 'CPU allocation ratio' is the maximum number of virtual processors that can be allocated to all cloud nodes (12, i.e., total processor resources) (e.g. in the case of Figure 3). , CPU allocation ratio means K*number of nodes).

At step S93, it may be determined whether the predicted overall utilization is less than the first threshold. For example, the provisioning system 10 may determine whether the overall utilization predicted for each of the processor resource and the memory resource is less than a first threshold (i.e., a first resource threshold and a second resource threshold). If it is less than the first threshold, step S95 may be performed, and vice versa, step S94 may be performed (i.e., an end target is additionally specified).

In step S94, it may be determined whether other spot instances exist in the pre-provisioned spot instance list. If other spot instances exist, steps S91 to S93 described above may be repeatedly performed for other spot instances. In the opposite case, provisioning for the requested regular instance may be determined to have failed.

In step S95, one or more spot instances designated as termination targets may be terminated.

Hereinafter, to provide easier understanding, a specific example of the spot instance termination method will be explained in detail with reference to FIG. 10.

Figure 10 assumes that the spot instance with the highest resource utilization rate is designated as the termination target, and that the first threshold consists of a first resource threshold for processor resources and a second resource threshold for memory resources.

As shown in FIG. 10, in step S101, a sorted list can be created by sorting the list of pre-provisioned spot instances based on the resource utilization rate of the instance. Specifically, a first sorted list sorted based on the usage rate of processor resources and a second sorted list sorted based on the usage rate of memory resources may be generated.

In step S102, the instance with the highest resource utilization in the sort list (e.g. the first instance when sorted in descending order) may be designated as a target for termination, and the designated instance may be removed from the sort list. If this step is performed repeatedly, the instance with the highest resource utilization can be alternately designated from the two sorted lists. For example, if the termination target instance was previously designated in the first sort list, the termination target may be specified in the second sort list this time, and the termination target may be specified in the first sort list again next time.

In step S103, the overall utilization rate of processor resources (usage rate _C , hereinafter referred to as 'first overall utilization rate') and the overall utilization rate of memory resources (utilization rate _R , hereinafter referred to as 'second overall utilization rate') of processor resources according to termination of the specified instance can be predicted, respectively. there is.

In step S104, it may be determined whether the first overall usage rate is less than the first resource threshold (TH1 _C ) or the second overall usage rate is less than the second resource threshold (TH1 _R ). Depending on the determination result, step S105 or step S106 may be performed.

In step S105, it may be determined whether other spot instances exist in the two sorted lists (or the sorted list for which a termination target will be specified next time). If present, steps S102 and S103 may be performed repeatedly for other spot instances (i.e., a termination target is additionally specified).

In step S106, an instance with the highest resource utilization may be additionally specified in the sorted list associated with the resource that does not meet the condition. And, the overall utilization of resources according to termination of additionally designated instances can be predicted again. For example, if the above condition is not met because the first overall utilization is more than the first resource threshold, a termination target may be additionally specified in the first sort list, and the total utilization of processor resources (utilization rate _C ) due to termination of the specified instance This can again be predicted.

In step S107, it may be determined again whether the predicted overall utilization is less than the resource threshold (eg TH1 _C ). Depending on the determination result, step S107 or step S108 may be performed. Figure 10 assumes a case where the overall utilization of processor resources is predicted again.

In step S108, it may be determined whether other spot instances exist in the two sorted lists (or the sorted list for which a termination target will be specified next time). If present, steps S106 and S107 may be performed repeatedly for other spot instances (i.e., a termination target is additionally specified).

In step S109, the specified instance may be terminated.

So far, a spot instance termination method according to an embodiment of the present disclosure has been described with reference to FIGS. 9 and 10 . According to the above, termination processing for Spot Instances can be performed in a simple but fair manner by determining which Spot Instances will be terminated first based on criteria such as utilization, bid price, and/or usage period. In addition, by predicting the overall utilization of cloud resources following termination of specified spot instances and terminating spot instances only until the predicted overall utilization falls below the first threshold, excessive termination of already provisioned spot instances can be prevented. You can.

Hereinafter, a spot instance provisioning method according to an embodiment of the present disclosure will be described with reference to FIGS. 11 and 12. However, for clarity of the present disclosure, description of content that overlaps with the previous embodiments will be omitted.

The provisioning method according to the embodiment may be performed in response to receiving a spot instance request from the client 13, or may be performed at the last point of the current time window (see FIG. 5). Hereinafter, the description will be continued assuming that the method is performed at the last point of the current time window.

Additionally, requests for Spot Instances include, as described above, the usage period of the instance, the required performance, the images to run on the instance, the bid price, the amount of processor resources, the amount of memory resources, and the acceptable latency (i.e., until provisioned). It may include information such as maximum waiting time).

Hereinafter, the provisioning method according to the embodiment will be described in more detail with reference to the exemplary flowchart shown in FIG. 11.

As shown in FIG. 11, the provisioning method according to the embodiment may also begin at step S111 where it is determined whether cloud resources are sufficient. If it is determined that the cloud resources are sufficient, step S112 may be performed, and in the opposite case, step S113 may be performed. For a description of this step, further refer to step S81 described above.

In step S112, it may be determined whether the overall utilization of cloud resources is less than the second threshold TH2. In response to determining that the second threshold is below, step S114 may be performed, and vice versa, step S113 may be performed. Here, the second threshold is a value that determines whether to provision a spot instance, and may be set to a value equal to or greater than the first threshold. Similar to the first threshold, the second threshold may also be set for each type of cloud resource. Also, the second threshold may be set to the same value or to different values for different resource types.

When the second threshold is set for each type of cloud resource, it may be determined whether the total usage rate of each cloud resource is less than the second threshold. For example, assume that the second threshold consists of a third resource threshold for processor resources and a fourth resource threshold for memory resources. In this case, the provisioning system 10 may determine whether the total utilization of processor resources is less than the third resource threshold and the total utilization of memory resources is also less than the fourth resource threshold. Then, the provisioning system 10 may perform step S114 in response to determining that both resources are below the resource threshold.

The second threshold may be a preset fixed value or a value that changes depending on the situation. For example, the second threshold may be a value that changes based on the frequency of occurrence of spot instance requests, the frequency of occurrence of regular instance requests, changes in cloud resource utilization, etc. As a more specific example, when the utilization rate of cloud resources changes rapidly or the frequency of regular instance requests increases rapidly, the provisioning system 10 reliably secures resources for regular instances and terminates already-provisioned spot instances excessively. To prevent this, the second threshold may be set to a smaller value. In the opposite case, provisioning system 10 may set the second threshold to a larger value. For example, if the frequency of regular instance requests decreases and the frequency of spot instance requests increases, the provisioning system 10 may set the second threshold to a larger value for more active provisioning of spot instances. Hereinafter, for convenience of understanding, 'second threshold', 'third resource threshold', and 'fourth resource threshold' will continue to be used with the same meaning as described above.

In order to provide easier understanding, the first and second thresholds will be briefly explained with reference to FIG. 12.

As shown in FIG. 12, the first threshold (TH1) may be a value that determines whether to terminate a previously provisioned spot instance, and the second threshold (TH2) may be a value that determines whether to provision a spot instance. As shown, the second threshold (TH2) can be set to a value greater than the first threshold (TH1), which utilizes the characteristics of spot instances (e.g., the usage period is not guaranteed and may be terminated prematurely). It can be understood as an attempt to improve the efficiency of using cloud resources.

Meanwhile, the difference between the first threshold TH1 and the second threshold TH2 may vary depending on the situation. For example, the difference between the two thresholds may be a value that changes based on the frequency of occurrence of spot instance requests, frequency of occurrence of regular instance requests, changes in cloud resource utilization, etc. As a more specific example, when the utilization rate of cloud resources changes rapidly or the frequency of regular instance requests increases rapidly, the provisioning system 10 reliably secures resources for regular instances and terminates already-provisioned spot instances excessively. To prevent this, the difference between the two thresholds can be set to be small. In the opposite case, the provisioning system 10 may set the difference between the two thresholds to become larger. For example, if the frequency of regular instance requests decreases and the frequency of spot instance requests increases, the provisioning system 10 may set the difference between the two thresholds to be larger for more active provisioning of spot instances.

Description will be made again with reference to FIG. 11 .

In step S113, provisioning for the requested spot instance may be determined to have failed.

In step S114, the number of spot instances that can be provisioned can be calculated. Specifically, the number of spot instances that can be provisioned may be calculated based on the number of virtual cloud resources (e.g. virtual processors, virtual memory) that can be allocated, the overall utilization rate of cloud resources, and/or a first threshold. However, specific calculation methods can be designed in various ways.

For example, assume that the number of allocable virtual processors is N _C-avail , the number of allocable virtual memories is N _R-avail , and the resource threshold for each processor resource and memory resource is set to '90%'. . At this time, N _c-avail may be calculated by subtracting the number of already allocated virtual processors from the maximum number of virtual processors that can be allocated to all cloud nodes 12. In this case, the provisioning system 10 may further calculate the assignable number (N _C , N _R ) based on the overall utilization of cloud resources according to Equations 1 and 2 below. Additionally, the provisioning system 10 may determine the minimum value among the calculated N _C-avail , N _R-avail , N _C , and N _R as the number of spot instances that can be provisioned.

In Equations 1 and 2 above, 'N _C ' means the allocable number of virtual processors, 'N _R ' means the allocable number of virtual memories, '90' means the threshold of each resource, and , 'AVG_10m_pCPU' and 'AVG_10m_pRAM' mean the overall utilization of processor resources and memory resources (eg, assumed to be the average value for 10 minutes of unit time), respectively. In addition, 'E_vCPU' and 'E_vRAM' refer to the predicted resource utilization of spot instances (i.e., predicted utilization of processor resources and predicted utilization of memory resources), and 'allocation ratio' refers to the previously mentioned 'CPU allocation ratio'. It can be understood to have the same or similar meaning as '. 'E_vCPU' and 'E_vRAM' can be calculated in any way. For example, 'E_vCPU' and 'E_vRAM' may be the average value of the resource utilization of previously provisioned spot instances, may be the resource utilization of a random spot instance, or may be values obtained from the client 13.

For reference, the numerator formula ('90-AVG_10m_pCPU') in Equation 1 can be understood as calculating the available ratio of processor resources by considering the resource threshold, and the denominator formula ('E_vCPU/allocation ratio') is the spot instance It can be understood as dividing the standard utilization rate by the 'allocation ratio' to convert it to the overall resource standard utilization rate. For example, if E_vCPU (i.e., processor resource utilization of the spot instance) is 64% and 'allocation ratio' is '16' (i.e., when up to 16 virtual processors can be assigned to all cloud nodes), the corresponding spot instance Since the utilization rate based on the total resources of the instance will only be '4%', it can be understood as dividing by the 'allocation ratio'.

In step S115, the requested spot instance may be provisioned. Specifically, if the number calculated in the previous step S114 is N, N spot instances may be provisioned. Additionally, when the number of requested spot instances exceeds N, N spot instances with relatively high bid prices may be provisioned.

In step S116, provisioning for the requested spot instance may be determined to be successful.

So far, a spot instance provisioning method according to an embodiment of the present disclosure has been described with reference to FIGS. 11 and 12. According to the above, the efficiency of using cloud resources can be improved by providing cloud services using spot instances. Additionally, by prohibiting provisioning of spot instances when the overall utilization of cloud resources is above the second threshold, the impact of spot instances on the performance of regular instances can be minimized, thereby ensuring the quality of service of regular instances. It can be. Furthermore, by accurately calculating the number of spot instances that can be provisioned, the efficiency of using cloud services can be further improved, and the profitability of cloud service providers can be improved by provisioning multiple spot instances using a price bidding method. .

Hereinafter, an exemplary computing device 130 capable of implementing the provisioning system 10 described above will be described.

13 is an exemplary hardware configuration diagram showing the computing device 130.

As shown in FIG. 13, the computing device 130 includes one or more processors 131, a bus 133, a communication interface 134, and a memory (loading) a computer program executed by the processor 131. 132) and a storage 135 that stores a computer program 136. However, only components related to the embodiment of the present disclosure are shown in FIG. 13. Accordingly, a person skilled in the art to which this disclosure pertains can see that other general-purpose components may be included in addition to the components shown in FIG. 13 . That is, the computing device 130 may further include various components in addition to those shown in FIG. 13 . Additionally, in some cases, the computing device 130 may be configured with some of the components shown in FIG. 13 omitted. Hereinafter, each component of the computing device 130 will be described.

The processor 131 may control the overall operation of each component of the computing device 130. The processor 131 is at least one of a Central Processing Unit (CPU), Micro Processor Unit (MPU), Micro Controller Unit (MCU), Graphic Processing Unit (GPU), or any type of processor well known in the art of the present disclosure. It can be configured to include. Additionally, the processor 131 may perform operations on at least one application or program to execute operations/methods according to embodiments of the present disclosure. Computing device 130 may include one or more processors.

Next, memory 132 may store various data, commands and/or information. Memory 132 may load a computer program 136 from storage 135 to execute operations/methods according to embodiments of the present disclosure. The memory 132 may be implemented as a volatile memory such as RAM, but the technical scope of the present disclosure is not limited thereto.

Next, the bus 133 may provide communication functionality between components of the computing device 130. The bus 133 may be implemented as various types of buses, such as an address bus, a data bus, and a control bus.

Next, the communication interface 134 may support wired or wireless Internet communication of the computing device 130. Additionally, the communication interface 134 may support various communication methods other than Internet communication. To this end, the communication interface 134 may be configured to include a communication module well known in the technical field of the present disclosure.

Next, storage 135 may non-transitory store one or more computer programs 136. The storage 135 may be a non-volatile memory such as Read Only Memory (ROM), Erasable Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), flash memory, a hard disk, a removable disk, or a device well known in the art to which this disclosure pertains. It may be configured to include any known type of computer-readable recording medium.

Next, the computer program 136 may include one or more instructions that, when loaded into the memory 132, cause the processor 131 to perform operations/methods according to various embodiments of the present disclosure. That is, the processor 131 may perform operations/methods according to various embodiments of the present disclosure by executing one or more loaded instructions.

For example, the computer program 136 may monitor utilization information including the overall utilization of cloud resources, receive a request for a regular instance from a client, and determine that the overall utilization is greater than or equal to a first threshold. In response, it may include instructions for terminating at least some of the already provisioned spot instances and provisioning the requested regular instances. In this case, the provisioning system 10 according to some embodiments of the present disclosure may be implemented through the computing device 130.

So far, an exemplary computing device 130 capable of implementing the provisioning system 10 according to some embodiments of the present disclosure has been described with reference to FIG. 13 .

So far, various embodiments of the present disclosure and effects according to the embodiments have been mentioned with reference to FIGS. 1 to 13 . The effects according to the technical idea of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description below.

In addition, although it has been described in the above embodiments that a plurality of components are combined or operated in combination, the technical idea of the present disclosure is not necessarily limited to these embodiments. That is, as long as it is within the scope of the technical idea of the present disclosure, all of the components may be operated by selectively combining one or more of them.

The technical ideas of the present disclosure described so far can be implemented as computer-readable code on a computer-readable medium. A computer program recorded on a computer-readable recording medium can be transmitted to another computing device through a network such as the Internet, installed on the other computing device, and thus used on the other computing device.

Although operations are shown in the drawings in a specific order, it should not be understood that the operations must be performed in the specific order shown or sequential order or that all illustrated operations must be performed to obtain the desired results. In certain situations, multitasking and parallel processing may be advantageous. Although various embodiments of the present disclosure have been described above with reference to the attached drawings, those skilled in the art will understand that the technical idea of the present disclosure can be modified in a different specific form without changing the technical idea or essential features. It is understandable that it can also be implemented. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. The scope of protection of this disclosure should be interpreted in accordance with the claims below, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of rights of the technical ideas defined by this disclosure.

Claims (17)

1. A method performed on at least one computing device, comprising:
monitoring utilization information including overall utilization of cloud resources;
Receiving a request for a regular instance from a client; and
In response to determining that the overall utilization is greater than or equal to a first threshold, terminating at least some of the already provisioned spot instances and provisioning the requested regular instances,
Provisioning methods for cloud services. According to paragraph 1,
The monitoring step is,
Monitoring resource utilization of the specific cloud node through a node agent operating on the specific cloud node providing the cloud resources; and
Comprising the step of monitoring resource utilization of the specific cloud instance through an instance agent operating on the specific cloud instance of the specific cloud node,
Provisioning methods for cloud services. According to paragraph 1,
The cloud resources include processor resources and memory resources,
The first threshold includes a first resource threshold for the processor resource and a second resource threshold for the memory resource,
The step of provisioning the requested regular instance is,
In response to determining that the total utilization of the processor resources is greater than or equal to the first resource threshold or that the total utilization of the memory resources is greater than or equal to the second resource threshold, terminating the at least some of the resources,
Provisioning methods for cloud services. According to paragraph 1,
The overall utilization rate of cloud resources is the first overall utilization rate,
The step of provisioning the requested regular instance is,
Specifying a termination target from the list of pre-provisioned spot instances;
predicting a second overall utilization of the cloud resources according to termination of the designated spot instance; and
In response to determining that the second overall utilization is below the first threshold, terminating the designated spot instance.
Provisioning methods for cloud services. According to clause 4,
The termination target is specified based on the resource utilization of the spot instance,
Provisioning methods for cloud services. According to clause 4,
The termination target is specified based on the bid price or used period of the spot instance,
Provisioning methods for cloud services. According to clause 4,
The step of provisioning the requested regular instance is,
In response to determining that the predicted total utilization is greater than or equal to the first threshold, further comprising specifying the termination target in the list of spot instances,
Provisioning methods for cloud services. According to clause 4,
The step of predicting the second overall utilization rate is,
Obtaining a utilization rate of virtual cloud resources allocated to the designated spot instance;
Predicting a decrease in overall resource utilization due to termination of the specified spot instance through a division operation between the obtained utilization rate and the maximum number of virtual cloud resources that can be allocated to all cloud nodes; and
Comprising the step of calculating the second overall utilization rate by reflecting the predicted decrease in the first overall utilization rate,
Provisioning methods for cloud services. According to paragraph 1,
Receiving a request for a spot instance from a client; and
In response to determining that the overall utilization is below a second threshold, provisioning the requested spot instance,
The second threshold is a value greater than the first threshold,
Provisioning methods for cloud services. According to paragraph 1,
Receiving requests for a plurality of spot instances from a client, the requests being received within a current time window; and
At the last point of the current time window, further comprising performing provisioning for the plurality of spot instances,
Provisioning methods for cloud services. According to clause 10,
The step of performing the provisioning is,
determining at least one spot instance among the plurality of spot instances based on a bid price; and
Including provisioning the determined spot instance,
Provisioning methods for cloud services. According to clause 10,
The step of performing the provisioning is,
Comprising the step of excluding instances that exceed the allowable waiting time of the corresponding client from among the plurality of spot instances from provisioning targets,
Provisioning methods for cloud services. According to clause 10,
The step of performing the provisioning is,
calculating the number of spot instances that can be provisioned based on the total utilization rate; and
Comprising the step of provisioning the calculated number of spot instances among the plurality of spot instances,
Provisioning methods for cloud services. According to clause 13,
The overall utilization includes a first overall utilization for processor resources and a second overall utilization for memory resources,
The step of calculating the number of spot instances that can be provisioned is,
calculating a first available ratio for the processor resource from the first total usage rate, and calculating a second available ratio for the memory resource from the second total usage rate;
calculating a first allocatable number for a virtual processor based on the first predicted resource utilization rate of the spot instance for the processor resource and the first available ratio;
calculating a second allocatable number for virtual memory based on a second predicted resource utilization rate of the spot instance for the memory resource and the second available ratio; and
Comprising the step of calculating the number of provisionable spot instances based on the first allocatable number and the second allocatable number,
Provisioning methods for cloud services. According to clause 14,
The first predicted resource utilization rate is,
Calculated through a division operation between the predicted utilization of the virtual processor assigned to the spot instance and the maximum number of virtual processors that can be assigned to all cloud nodes,
Provisioning methods for cloud services. One or more processors; and
Includes memory for storing one or more instructions,
The one or more processors:
By executing one or more instructions stored above,
Monitoring utilization information, including overall utilization of cloud resources;
An operation of receiving a request for a regular instance from a client, and
In response to determining that the overall utilization is greater than or equal to a first threshold, performing an operation of terminating at least some of the already provisioned spot instances and provisioning the requested regular instances.
Provisioning system. Combined with a computing device,
monitoring utilization information including overall utilization of cloud resources;
Receiving a request for a regular instance from a client; and
A computer-readable record for executing the step of terminating at least a portion of already provisioned spot instances and provisioning the requested regular instances, in response to determining that the overall utilization is greater than or equal to a first threshold. stored on the medium,
computer program.

Images