Container management based on application performance indicators

Patent metadata

Field	Value
Publication number	US-11301288-B2
Application number	US-202016870105-A
Country	US
Kind code	B2
Filing date	May 8, 2020
Priority date	May 8, 2020
Publication date	Apr 12, 2022
Grant date	Apr 12, 2022

Abstract

Techniques for managing containers based on application key performance indicators (KPIs), associated with instances of network applications executing within containers in a telecommunication network, are discussed herein. For example, a container manager can determine that an instance of the network application is underperforming a target KPI value, is otherwise experiencing problems, and/or may be likely to experience future problems. The container manager can accordingly take one or more corrective and/or preventative actions, such as to terminate and replace the container associated with the underperforming instance of the network application, or to scale out a set of containers by adding additional containers and corresponding additional instances of the network application to reduce the load on individual instances of the network application.

First claim

What is claimed is:

1. A method of managing a container in a telecommunication network, comprising:
receiving, by a container manager, an application key performance indicator (KPI) associated with an instance of a network application executing within the container;
determining, by the container manager, a KPI difference between a target KPI value and the application KPI associated with the instance of the network application executing within the container;
determining, by the container manager, that the KPI difference exceeds a threshold difference; and
performing, by the container manager, at least one corrective action associated with the container based on determining that the KPI difference exceeds the threshold difference;
training, by the container manager, a machine learning model based on correlations between one or more container KPIs from the set of containers and one or more application KPIs including the application KPI; and
predicting, by the container manager using the machine learning model, one or more subsequent application KPIs associated with the instance of the network application based on subsequent container KPIs from the set of containers.

2. The method of claim 1 , wherein the at least one corrective action comprises:
terminating, by the container manager, the container executing the instance of the network application; and
creating, by the container manager, a replacement container that executes a replacement instance of the network application.

3. The method of claim 1 , wherein the at least one corrective action comprises adding, by the container manager, at least one additional container to a set of containers, wherein the at least one additional container executes at least one additional instance of the network application.

4. The method of claim 1 , wherein the at least one corrective action comprises at least one of:
terminating and replacing, by the container manager and based at least in part on the KPI difference exceeding a first threshold, the container in a set of containers managed by the container manager; or
adding, by the container manager and based at least in part on the KPI difference exceeding a second threshold that is lower than the first threshold, an additional container to the set of containers.

5. The method of claim 1 ,
wherein the application KPI associated with the instance of the network application is one of the one or more subsequent application KPIs predicted for the instance of the network application.

6. The method of claim 1 , further comprising deriving, by the container manager, the target KPI value based on statistical data associated with values of the application KPI associated with a set of instances of the network application executing within a set of containers.

7. The method of claim 1 , wherein the network application is a fifth generation (5G) network function.

8. The method of claim 7 , wherein the 5G network function comprises at least one of a Session Management Function (SMF), a User Plane Function (UPF), or an Access and Mobility Management Function (AMF).

9. The method of claim 1 , wherein the application KPI includes at least one of:
a Protocol Data Unit (PDU) activation success rate;
a handover success rate;
a request timeout;
an application level packet drop rate;
a session average throughput metric;
a number of failed sessions;
call latency information;
dropped call information; or
a number of messages rejected by other instances of network applications in other containers.

10. A container manager, comprising:
one or more processors; and
memory storing computer-executable instructions that, when executed by the one or more processors, cause the container manager to perform operations comprising:
receiving application key performance indicators (KPIs) associated with a set of instances of a network application executing within a set of containers;
determining a KPI difference between a target KPI value and first application KPIs associated with a first instance of the network application executing within a first container in the set of containers;
determining that the KPI difference exceeds a threshold difference; and
performing at least one corrective action on the set of containers based on determining that the KPI difference exceeds the threshold difference,

wherein an application KPI associated with the first instance of the network application is received by the container manager from a second instance of a second network application executing within a second container, and
wherein the network application is a first type of fifth generation (5G) network function and the second network application is a second type of 5G network function.

11. The container manager of claim 10 , wherein the at least one corrective action comprises at least one of:
terminating the first container executing the first instance of the network application and creating a replacement container in the set of containers that executes a replacement instance of the network application, or
adding at least one additional container to the set of containers that executes at least one additional instance of the network application.

12. The container manager of claim 10 , wherein the operations further comprise:
receiving container KPIs from the set of containers;
training a machine learning model based on correlations between the container KPIs and the application KPIs;
receiving subsequent container KPIs from the set of containers; and
predicting, using the machine learning model, subsequent application KPIs associated with the set of instances of the network application based on the subsequent container KPIs,
wherein the first application KPIs associated with the first instance of the network application are the subsequent application KPIs predicted for the first instance of the network application.

13. The container manager of claim 10 , wherein the operations further comprise deriving the target KPI value based on statistical data associated with values of the application KPIs associated with the set of instances of the network application executing within the set of containers.

14. The container manager of claim 10 , wherein the application KPIs include at least one of:
Protocol Data Unit (PDU) activation success rates;
handover success rates;
request timeouts;
application level packet drop rates;
session average throughput metrics;
a number of failed sessions;
call latency information;
dropped call information; or
a number of messages rejected by other instances of network applications in other containers.

15. One or more non-transitory computer-readable media storing computer-executable instructions that, when executed by one or more processors of at least one computing device of a telecommunication network, cause the at least one computing device to perform operations comprising:
receiving application key performance indicators (KPIs) associated with a set of instances of a network application executing within a set of containers;
determining a KPI difference between a target KPI value and first application KPIs associated with a first instance of the network application executing within a first container in the set of containers;
determining that the KPI difference exceeds a threshold difference; and
performing at least one corrective action on the set of containers based on determining that the KPI difference exceeds the threshold difference,
training a machine learning model based on correlations between container KPIs from the set of containers and the application KPIs; and
predi

Assignees

• T Mobile Usa Inc

Inventors

• Albasheir Suliman

Classifications

• G06F11/302

  where the computing system component is a software system · CPC title

• G06F11/3006

  where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems (multiprogramming arrangements G06F9/46; allocation of resources G06F9/50) · CPC title

• H04L67/1004 Primary

  Server selection for load balancing · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F11/3409

  for performance assessment · CPC title