Granted Patent B2

US 12,737,238 · App. 18/195,620 · Granted Sep 15, 2026

Inventors: Hyo Jung Lee (Seoul, KR); Thanh-Tung Nguyen (Seoul, KR); Thanh-Cuong Nguyen (Seoul, KR); Viet-Anh Nguyen (Seoul, KR); Thi-Thanh-Lam Vu (Seoul, KR)

Assignee: SAMSUNG SDS CO., LTD.

G06F9/5088

Abstract

Claims (69)

1 . A provisioning method performed by at least one computing device for a cloud service, the provisioning method comprising:

collecting resource utilization rates of individual cloud nodes of a plurality of cloud nodes, wherein a resource utilization rate of an individual cloud node includes a resource utilization rate for at least one of a processor or a memory of the individual cloud node;

receiving a regular instance request from at least one client;

determining whether there is a cloud node among the plurality of cloud nodes that satisfies a first condition, in which the collected resource utilization rate is less than a first threshold value;

designating a specific cloud node from among the plurality of cloud nodes based on a determination that there is no cloud node among the plurality of cloud nodes that satisfies the first condition;

terminating at least some of spot instances pre-provisioned and operating on the specific cloud node and provisioning a requested regular instance in the specific cloud node;

receiving, from the at least one client, a plurality of spot instance requests within a current time window;

determining at least one cloud node among the plurality of cloud nodes that satisfies a second condition, in which the collected resource utilization rate is less than a second threshold value; and

performing, at a last point of the current time window, provisioning for requested plurality of spot instances based on a resource utilization rate of the determined at least one cloud node.

2 . The provisioning method of claim 1 , wherein the collecting the resource utilization rate comprises:

collecting a resource utilization rate of the specific cloud node through a node agent operating on the specific cloud node; and

collecting a resource utilization rate of a specific cloud instance through an instance agent operating on the specific cloud instance of the specific cloud node.

3 . The provisioning method of claim 1 , wherein the provisioning the requested regular instance comprises:

designating a termination target from a list of the pre-provisioned spot instances;

predicting a resource utilization rate of the specific cloud node according to termination of designated spot instance; and

terminating the designated spot instance based on a determination that the predicted resource utilization rate is less than the first threshold value.

4 . The provisioning method of claim 3 , wherein the termination target is designated based on a resource utilization rate of a spot instance.

5 . The provisioning method of claim 3 , wherein the termination target is designated based on at least one of a bid price or a used period of a spot instance.

6 . The provisioning method of claim 3 , wherein the provisioning the requested regular instance further comprises:

further designating another termination target from the list of the pre-provisioned spot instances based on a determination that the predicted resource utilization rate is equal to or greater than the first threshold value.

7 . The provisioning method of claim 1 ,

wherein the second threshold value is set to a value greater than the first threshold value.

8 . The provisioning method of claim 1 , wherein the performing the provisioning for the requested plurality of spot instances comprises:

excluding, from a provisioning target, an instance that exceeds an allowable waiting time of a corresponding client among the requested plurality of spot instances.

9 . The provisioning method of claim 1 , wherein the performing the provisioning for the requested plurality of spot instances comprises:

determining at least one spot instance among the requested plurality of spot instances based on a bid price; and

provisioning the at least one determined spot instance.

10 . The provisioning method of claim 1 , wherein the performing the provisioning for the requested plurality of spot instances comprises:

calculating a total number of spot instances that can be provisioned in the determined at least one cloud node based on a the resource utilization rate of the determined at least one cloud node; and

provisioning the calculated total number of spot instances among the requested plurality of spot instances.

11 . The provisioning method of claim 10 , wherein the calculating the total number of spot instances that can be provisioned comprises:

calculating a resource availability rate of a first cloud node, among the determined at least one cloud node, based on a resource utilization rate of the first cloud node and the second threshold value;

calculating a representative resource utilization rate for a spot instance based on a resource utilization rate of a spot instance operating in the first cloud node; and

calculating a number of spot instances that can be provisioned in the first cloud node based on the resource availability rate and the representative resource utilization rate.

12 . A provisioning method performed by at least one computing device for a cloud service, the provisioning method comprising:

collecting resource utilization rates of individual cloud nodes of a plurality of cloud nodes, wherein a resource utilization rate of an individual cloud node includes a resource utilization rate for at least one of a processor or a memory of the individual cloud node;

receiving a regular instance request from at least one client;

determining whether there is a cloud node among the plurality of cloud nodes that satisfies a first condition, in which the collected resource utilization rate is less than a first threshold value;

determining a source cloud node from among the plurality of cloud nodes based on a determination that there is no cloud node among the plurality of cloud nodes that satisfies the first condition;

migrating at least some of spot instances pre-provisioned and operating on the source cloud node to a target cloud node;

provisioning a requested regular instance at the source cloud node;

receiving, from the at least one client, a plurality of spot instance requests within a current time window;

determining at least one cloud node among the plurality of cloud nodes that satisfies a second condition, in which the collected resource utilization rate is less than a second threshold value; and

performing, at a last point of the current time window, provisioning for the requested plurality of spot instances based on a resource utilization rate of the determined at least one cloud node.

13 . The provisioning method of claim 12 , wherein the determining the source cloud node comprises:

determining whether a first cloud node among the plurality of cloud nodes satisfies the first condition according to spot instance migration or termination; and

determining the first cloud node as the source cloud node based on a determination that the first cloud node satisfies the first condition.

14 . The provisioning method of claim 13 , wherein the determining the source cloud node further comprises:

determining, based on the first cloud node not satisfying the first condition, whether a second cloud node among the plurality of cloud nodes satisfies the first condition according to the spot instance migration or the termination.

15 . The provisioning method of claim 12 , wherein the migrating the at least some of the spot instances of the source cloud node comprises:

determining the target cloud node from among the plurality of cloud nodes based on a number of spot instances that can be provisioned in the target cloud node; and

migrating the at least some of spot instances of the source cloud node to the determined target cloud node.

16 . The provisioning method of claim 15 , wherein the migrating to the determined target cloud node comprises:

migrating spot instances of the source cloud node as many as the number of spot instances that can be provisioned in the determined target cloud node;

wherein spot instances remaining on the source cloud node are terminated before the requested regular instance is provisioned.

17 . The provisioning method of claim 12 ,

wherein the second threshold value is set to a value greater than the first threshold value.

18 . A provisioning system for a cloud service, the provisioning system comprising:

at least one processor; and

at least one memory configured to store computer program executable by the at least one processor,

wherein the computer program, when executed by the at least one processor, causes the at least one processor to perform:

collecting resource utilization rates of individual cloud nodes of a plurality of cloud nodes, wherein a resource utilization rate of an individual cloud node includes a resource utilization rate for at least one of a processor or a memory of the individual cloud node;

receiving a regular instance request from at least one client;

determining whether there is a cloud node among the plurality of cloud nodes that satisfies a first condition, in which the collected resource utilization rate is less than a first threshold value;

designating a specific cloud node from among the plurality of cloud nodes based on a determination that there is no cloud node among the plurality of cloud nodes that satisfies the first condition;

terminating at least some of spot instances pre-provisioned and operating on the specific cloud node and provisioning a requested regular instance in the specific cloud node;

receiving, from the at least one client, a plurality of spot instance requests within a current time window;

determining at least one cloud node among the plurality of cloud nodes that satisfies a second condition, in which the collected resource utilization rate is less than a second threshold value; and

performing, at a last point of the current time window, provisioning for the requested plurality of spot instances based on a resource utilization rate of the determined at least one cloud node.

Assignments (1)

Priority Claims (3)

Continuity (1)

References Cited (18)