Intelligent auto-scaling of containerized workloads in container computing environment
Techniques for managing containerized workloads in a container computing environment are disclosed. For example, a method comprises the following steps. The method predicts a composite time delay value for initializing an instance of a containerized workload for executing a microservice within a container computing environment. The method then computes at least one target resource utilization parameter, based on the predicted composite time delay value, for use by the container computing environment.
1 . An apparatus comprising:
at least one processing platform comprising at least one processor coupled to at least one memory, the at least one processing platform, when executing program code, is configured to:
predict, using a prediction model, a composite time delay value for initializing an instance of a containerized workload for executing a microservice of a given microservice-based application within a container computing environment during a given time interval in a given cluster of the container computing environment based on an input of one or more cluster-dependent factors associated with the given cluster and one or more application-dependent factors associated with the given microservice-based application, wherein the microservice is indicated as having one of a first level of criticality and a second level of criticality, and wherein the composite time delay value is calculated based on whether the microservice is indicated as having one of the first level of criticality and the second level of criticality;
compute at least one dynamic target resource utilization parameter, based on the predicted composite time delay value, for use by the given cluster of the container computing environment;
iteratively re-predict, using the prediction model, the composite time delay value at another time interval to measure and record a predicted rate of change for the containerized workload for executing the microservice of the given microservice-based application based at least on the predicted composite time delay value for the given time interval and the re-predicted composite time delay value for the other time interval;
automatically adjust the at least one dynamic target resource utilization parameter based on the re-predicted composite time delay value and the predicted rate of change; and
initialize an additional instance of the containerized workload based on the automatically adjusted at least one dynamic target resource utilization parameter.
2 . The apparatus of claim 1 , wherein predicting the composite time delay value further comprises computing a time delay factor associated with detection of a condition for initiating a new containerized workload instance.
3 . The apparatus of claim 2 , wherein a maximum value for the time delay factor associated with detection of a condition for initiating a new containerized workload instance comprises a sum of a sync time component and a metric resolution component.
4 . The apparatus of claim 3 , wherein a minimum value for the time delay factor associated with detection of a condition for initiating a new containerized workload instance comprises the metric resolution component.
5 . The apparatus of claim 4 , wherein predicting the composite time delay value further comprises computing a time delay factor associated with starting an application program to be executed in a new containerized workload instance.
6 . The apparatus of claim 5 , wherein the time delay factor associated with starting an application program to be executed in a new containerized workload instance comprises an image download time component and a container initialization time component.
7 . The apparatus of claim 6 , wherein the image download time component is computed by applying one or more machine learning-based algorithms to image size data and resource utilization data in multiple time intervals.
8 . The apparatus of claim 6 , wherein the container initialization time component is computed based on a default value or a custom value.
9 . The apparatus of claim 8 , wherein predicting a maximum value for the composite time delay value further comprises summing the sync time component, the metric resolution component, the image download time component, and the container initialization time component.
10 . The apparatus of claim 9 , wherein predicting a minimum value for the composite time delay value further comprises summing the metric resolution component, the image download time component, and the container initialization time component.
11 . The apparatus of claim 10 , wherein when the microservice is identified as having the first level of criticality, the maximum value of the composite time delay value is used to compute the at least one dynamic target resource utilization parameter.
12 . The apparatus of claim 10 , wherein when the microservice is identified as having the second level of criticality, the mean of the maximum value and the minimum value of the composite time delay value is used to compute the at least one dynamic target resource utilization parameter.
13 . A method comprising:
predicting, using a prediction model, a composite time delay value for initializing an instance of a containerized workload for executing a microservice of a given microservice-based application within a container computing environment during a given time interval in a given cluster of the container computing environment based on an input of one or more cluster-dependent factors associated with the given cluster and one or more application-dependent factors associated with the given microservice-based application, wherein the microservice is indicated as having one of a first level of criticality and a second level of criticality, and wherein the composite time delay value is calculated based on whether the microservice is indicated as having one of the first level of criticality and the second level of criticality, and wherein the composite time delay value is calculated based on whether the microservice is indicated as having one of the first level of criticality and the second level of criticality;
computing at least one dynamic target resource utilization parameter, based on the predicted composite time delay value, for use by the given cluster of the container computing environment;
iteratively re-predicting, using the prediction model, the composite time delay value at another time interval to measure and record a predicted rate of change for the containerized workload for executing the microservice of the given microservice-based application based at least on the predicted composite time delay value for the given time interval and the re-predicted composite time delay value for the other time interval;
automatically adjusting the at least one dynamic target resource utilization parameter based on the re-predicted composite time delay value and the predicted rate of change; and
initializing an additional instance of the containerized workload based on the automatically adjusted at least one dynamic target resource utilization parameter;
wherein the method is performed by at least one processing device comprising a processor coupled to a memory when executing program code.
14 . The method of claim 13 , wherein predicting the composite time delay value further comprises:
computing a time delay factor associated with detection of a condition for initiating a new containerized workload instance; and
computing a time delay factor associated with starting an application program to be executed in a new containerized workload instance.
15 . The method of claim 14 , wherein:
a maximum value for the time delay factor associated with detection of a condition for initiating a new containerized workload instance comprises a sum of a sync time component and a metric resolution component; and
a minimum value for the time delay factor associated with detection of a condition for initiating a new containerized workload instance comprises the metric resolution component.
16 . The method of claim 15 , wherein the time delay factor associated with starting an application program to be executed in a new containerized workload instance comprises an image download time component and a container initialization time component.
17 . The method of claim 16 , wherein:
predicting a maximum value for the composite time delay value further comprises summing the sync time component, the metric resolution component, the image download time component, and the container initialization time component; and
predicting a minimum value for the composite time delay value further comprises summing the metric resolution component, the image download time component, and the container initialization time component.
18 . The method of claim 17 , wherein:
when the microservice is identified as having the first level of criticality, the maximum value of the composite time delay value is used to compute the at least one dynamic target resource utilization parameter; and
when the microservice is identified as having the second level of criticality, the mean of the maximum value and the minimum value of the composite time delay value is used to compute the at least one dynamic target resource utilization parameter.
19 . A computer program product comprising a non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing platform causes the at least one processing platform to:
predict, using a prediction model, a composite time delay value for initializing an instance of a containerized workload for executing a microservice of a given microservice-based application within a container computing environment during a given time interval in a given cluster of the container computing environment based on an input of one or more cluster-dependent factors associated with the given cluster and one or more application-dependent factors associated with the given microservice-based application, wherein the microservice is indicated as having one of a first level of criticality and a second level of criticality, and wherein the composite time delay value is calculated based on whether the microservice is indicated as having one of the first level of criticality and the second level of criticality, and wherein the composite time delay value is calculated based on whether the microservice is indicated as having one of the first level of criticality and the second level of criticality;
compute at least one dynamic target resource utilization parameter, based on the predicted composite time delay value, for use by the given cluster of the container computing environment;
iteratively re-predict, using the prediction model, the composite time delay value at another time interval to measure and record a predicted rate of change for the containerized workload for executing the microservice of the given microservice-based application based at least on the predicted composite time delay value for the given time interval and the re-predicted composite time delay value for the other time interval;
automatically adjust the at least one dynamic target resource utilization parameter based on the re-predicted composite time delay value and the predicted rate of change; and
initialize an additional instance of the containerized workload based on the automatically adjusted at least one dynamic target resource utilization parameter.
20 . The computer program product of claim 19 , wherein predicting the composite time delay value further comprises:
computing a time delay factor associated with detection of a condition for initiating a new containerized workload instance; and
computing a time delay factor associated with starting an application program to be executed in a new containerized workload instance.