Method for rapid service deployment in hybrid cloud environment
There are provided a method and an apparatus for hybrid cloud management, which configure, reconfigure, and manage service resources in order to rapidly deploy a service in a hybrid cloud environment. According to embodiments of the disclosure, when there is a request for resources of a service operating in an existing cloud environment (Kubernetes), problems of a method of simply expanding replicas may be solved, and rapid processing (deployment) may be performed in response to a continuous resource request. In addition, an available space for using resources may be guaranteed by applying a method of HPA (increasing the number of resource replicas), VPA (increasing allocated resources), migration (transferring resources), rather than simply expanding the number of replicas.
1 . A processor-implemented method for hybrid cloud management, the method comprising:
collecting, by a hybrid cloud management device, metric data regarding a current resource use rate of one or more computing resources associated with an application service from a plurality of nodes included in one or more clusters supporting the application service operating in a hybrid cloud environment;
determining, by the hybrid cloud management device, the current resource use rate, based on the collected metric data;
performing resource defragmentation by allocating available resources to the one or more clusters according to a pre-set policy used in a policy-based threshold determination, the policy-based threshold determination including determining whether the current resource use rate satisfies a pre-set threshold value included in the pre-set policy, based on the resource use rate; and
providing a requested service executed in a node of a cluster, from among plural nodes, based on the resource defragmentation,
wherein the performing resource defragmentation further includes:
predicting a predicted resource use rate of a future time regarding a plurality of nodes included in one or more interlocking clusters by using a prediction model;
filtering a node whose predicted resource use rate increases by the pre-set threshold value or more, based on a result of the predicting;
selecting, as a migration target cluster, a cluster that includes a node having a lowest predicted resource use rate among remaining nodes after the filtering; and
performing migration of the filtered node to the migration target cluster depending on a determination that the predicted resource use rate remains greater than the pre-set threshold value even after performing at least one of vertical pod autoscaling and horizontal pod autoscaling.
2 . The method of claim 1 , wherein the performing the resource defragmentation comprises, when the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy, performing the resource defragmentation for guaranteeing the resources.
3 . The method of claim 2 , wherein the performing the resource defragmentation comprises performing the resource defragmentation for guaranteeing the resources with respect to all clouds supporting the application service operating in the hybrid cloud environment, and
wherein the resource defragmentation is performed by performing at least two methods of Horizontal Pod Autoscaling (HPA) by increasing a number of resource replicas, Vertical Pod Autoscaling (VPA) by increasing allocated resources allocated for the application service, and migration by transferring resources, simultaneously, rather than simply expanding a number of replicas.
4 . The method of claim 3 , wherein the performing the resource defragmentation comprises:
scaling using at least one method of the VPA and the HPA by considering a current state of available resources to be allocated for the application service; and
when the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy even when scaling is performed, performing migration.
5 . The method of claim 4 , wherein the hybrid cloud management device comprises a processor comprising:
a policy management module configured to manage a policy set by a user;
a policy check module configured to acquire information regarding the set policy from the policy management module, and to determine whether the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy, based on the information regarding the set policy; and
a resource management module configured to perform resource defragmentation when the current resource use rate is greater than or equal to the threshold value included in the pre-set policy.
6 . The method of claim 5 , wherein the resource management module is configured to, when available resources of a node, the node including plural pods, are less than resources allocated to at least one pod of the plural pods, scale the resources of the at least one pod in the VPA by increasing the resources allocated to the at least one pod.
7 . The method of claim 5 , wherein the resource management module is configured to, when available resources of a node including a specific pod are larger than or equal to resources allocated to the specific pod, scale the resources of the specific pod in the HPA by generating a replica pod in the corresponding node by copying the replica pod as the specific pod, and by setting the specific pod and the replica pod to perform a same service.
8 . The method of claim 5 , wherein the resource management module is configured to perform migration between clusters when the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy even after scaling is performed by at least one method of the VPA and the HPA.
9 . A non-transitory computer-readable recording medium having a computer program recorded thereon to configure a processor to perform a hybrid cloud management method, the method comprising:
collecting, by a hybrid cloud management device, metric data regarding a current resource use rate of one or more computing resources associated with an application service from a plurality of nodes included in one or more clusters supporting the application service operating in a hybrid cloud environment; and
determining, by the hybrid cloud management device, the current resource use rate, based on the collected metric data;
performing resource defragmentation by allocating available resources to the one or more clusters according to a pre-set policy used in a policy-based threshold determination, the policy-based threshold determination including determining whether the current resource use rate satisfies the pre-set threshold value included in the pre-set policy, based on the resource use rate; and
providing a requested service executed in a node of a cluster, from among plural nodes, based on the resource defragmentation,
wherein the performing resource defragmentation further includes:
predicting a predicted resource use rate of a future time regarding a plurality of nodes included in one or more interlocking clusters by using a prediction model;
filtering a node whose predicted resource use rate increases by the pre-set threshold value or more, based on a result of the predicting;
selecting, as a migration target cluster, a cluster that includes a node having a lowest predicted resource use rate among remaining nodes after the filtering; and
performing migration of the filtered node to the migration target cluster depending on a determination that the predicted resource use rate remains greater than the pre-set threshold value even after performing at least one of vertical pod autoscaling and horizontal pod autoscaling.
10 . A hybrid cloud management device comprising:
a processor configured to:
collect metric data regarding a current resource use rate of one or more computing resources associated with an application service from a plurality of nodes included in one or more clusters supporting an application service operating in a hybrid cloud environment;
determine the current resource use rate, based on the collected metric data;
perform resource defragmentation by allocating available resources to the one or more clusters according to a pre-set policy used in a policy-based threshold determination, the policy-based threshold determination including determining whether the current resource use rate satisfies a pre-set threshold value included in the pre-set policy, based on the resource use rate; and
provide a requested service executed in a node of a cluster, from among plural nodes, based on the resource defragmentation,
wherein the performing resource defragmentation further includes:
predicting a predicted resource use rate of a future time regarding a plurality of nodes included in one or more interlocking clusters by using a prediction model;
filtering a node whose predicted resource use rate increases by the pre-set threshold value or more, based on a result of the predicting;
selecting, as a migration target cluster, a cluster that includes a node having a lowest predicted resource use rate among remaining nodes after the filtering; and
performing migration of the filtered node to the migration target cluster depending on a determination that the predicted resource use rate remains greater than the pre-set threshold value even after performing at least one of vertical pod autoscaling and horizontal pod autoscaling.
11 . The device of claim 10 , wherein the performing the resource defragmentation comprises, when the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy, performing the resource defragmentation for guaranteeing the resources.
12 . The device of claim 11 , wherein the performing the resource defragmentation comprises performing the resource defragmentation for guaranteeing the resources with respect to all clouds supporting the application service operating in the hybrid cloud environment, and
wherein the resource defragmentation is performed by performing at least two methods of Horizontal Pod Autoscaling (HPA) by increasing a number of resource replicas, Vertical Pod Autoscaling (VPA) by increasing allocated resources allocated for the application service, and migration by transferring resources, simultaneously, rather than simply expanding a number of replicas.
13 . The device of claim 12 , wherein the performing the resource defragmentation comprises:
scaling using at least one method of the VPA and the HPA by considering the current state of available resources to be allocated for the application service; and
when the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy even when scaling is performed, performing migration.
14 . The device of claim 13 , wherein the processor further comprising:
a policy management module configured to manage a policy set by a user;
a policy check module configured to acquire information regarding the set policy from the policy management module, and to determine whether the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy, based on the information regarding the set policy; and
a resource management module configured to perform resource defragmentation when the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy.
15 . The device of claim 14 , wherein the resource management module is configured to, when available resources of a node, the node including plural pods, are less than resources allocated to at least one pod of the plural pods, scale the resources of the at least one pod in the VPA by increasing the resources allocated to the at least one pod.
16 . The device of claim 14 , wherein the resource management module is configured to, when available resources of a node including a specific pod are larger than or equal to resources allocated to the specific pod, scale the resources of the specific pod in the HPA by generating a replica pod in the corresponding node by copying the replica pod as the specific pod, and by setting the specific pod and the replica pod to perform a same service.
17 . The device of claim 14 , wherein the resource management module is configured to perform migration between clusters when the current resource use rate is greater than or equal to the pre-set threshold value included in the pre-set policy even after scaling is performed by at least one method of the VPA and the HPA.