Automatically managing performance of software in a distributed computing environment

1. A system comprising:

a processor; and

a memory including instructions that are executable by the processor for causing the processor to:

deploy a first instance of a service in a distributed computing environment using a definition file configured for deploying instances of the service;

receive metrics information describing resource usage by the first instance of the service in the distributed computing environment;

determine a quality-of-service (QoS) constraint for the service; and

subsequent to receiving the metrics information and determining the QoS constraint:

automatically modify the definition file based on the metrics information and the QoS constraint; and

deploy a second instance of the service in the distributed computing environment using the modified definition file, the second instance being configured to more closely satisfy the QoS constraint than the first instance as a result of the modification to the definition file.

2. The system of claim 1 , wherein the metrics information includes resource-usage metrics describing memory usage, disk usage, processing usage, or network usage information of the first instance.

3. The system of claim 2 , wherein the metrics information also includes QoS metrics describing a latency, a responsiveness, an availability, or a reliability of the first instance.

4. The system of claim 1 , wherein the memory further includes instructions that are executable by the processor for causing the processor to determine the QoS constraint by obtaining the QoS constraint from the definition file.

5. The system of claim 4 , wherein the QoS constraint is predefined in the definition file by a user.

6. The system of claim 1 , wherein the memory further includes instructions that are executable by the processor for causing the processor to:

provide the metrics information and the QoS constraint as input to a trained machine-learning model for receiving tuning information as output from the trained machine-learning model, the tuning information indicating an adjustment to a resource specification described in the definition file for the service, the adjustment being for causing the second instance of the service to more closely satisfy the QoS constraint than the first instance; and

modify the resource specification in the definition file based on the tuning information to generate the modified definition file.

7. The system of claim 6 , wherein the resource specification indicates a minimum amount of a computing resource to allocate to the service in the distributed computing environment.

8. The system of claim 1 , wherein the memory further includes instructions that are executable by the processor for causing the processor to iteratively perform a tuning process involving (i) deploying a respective instance of the service in the distributed computing environment using the definition file, (ii) receiving respective metrics information corresponding to the respective instance of the service, and (iii) modifying one or more aspects of the definition file based on the respective metrics information and the QoS constraint, wherein the tuning process is configured to cause at least one later instance of the service deployed later in time to more closely satisfy the QoS constraint than at least one earlier instance of the service deployed earlier in time.

9. The system of claim 1 , wherein the memory further includes instructions that are executable by the processor for causing the processor to:

shut down the first instance in response to deploying the second instance; and

route client requests to the second instance subsequent to shutting down the first instance.

10. The system of claim 1 , wherein the memory further includes instructions that are executable by the processor for causing the processor to deploy the first instance and the second instance at a physical edge of the distributed computing environment, and wherein the service includes a serverless function.

11. A method comprising:

deploy, by one or more processors, a first instance of a service in a distributed computing environment using a definition file configured for deploying instances of the service;

receiving, by the one or more processor, metrics information describing resource usage by the first instance of the service in the distributed computing environment;

determining, by the one or more processors, a quality-of-service (QoS) constraint for the service;

subsequent to receiving the metrics information and determining the QoS constraint:

automatically modifying, by the one or more processors, the definition file based on the metrics information and the QoS constraint, the definition file being configured for deploying instances of the service in the distributed computing environment; and

deploying, by the one or more processors, a second instance of the service in the distributed computing environment using the modified definition file, the second instance being configured to more closely satisfy the QoS constraint than the first instance as a result of the modification to the definition file.

12. The method of claim 11 , wherein the metrics information includes resource-usage information describing memory usage, disk usage, processing usage, or network usage of the first instance.

13. The method of claim 11 , wherein the QoS constraint includes a latency constraint, a responsiveness constraint, an availability constraint, or a reliability constraint.

14. The method of claim 11 , further comprising determining the QoS constraint by obtaining the QoS constraint from the definition file.

15. The method of claim 11 , further comprising:

providing the metrics information and the QoS constraint as input to a trained machine-learning model for receiving tuning information as output from the trained machine-learning model, the tuning information indicating an adjustment to a portion of the definition file for the service, the adjustment being for causing the second instance of the service to more closely satisfy the QoS constraint than the first instance; and

modifying the portion of the definition file based on the tuning information to generate the modified definition file.

16. The method of claim 15 , wherein the portion of the definition file is a resource specification indicating how computing resources are to be allocated to the service.

17. The method of claim 11 , further comprising iteratively performing a tuning process involving (i) deploying a respective instance of the service in the distributed computing environment using the definition file, (ii) receiving respective metrics information corresponding to the respective instance of the service, and (iii) modifying one or more aspects of the definition file based on the respective metrics information and the QoS constraint, wherein iteratively performing the tuning process causes at least one later instance of the service deployed later in time to more closely satisfy the QoS constraint than at least one earlier instance of the service deployed earlier in time.

18. The method of claim 11 , further comprising:

shutting down the first instance in response to deploying the second instance; and

routing client requests to the second instance subsequent to shutting down the first instance.

19. The method of claim 11 , further comprising deploying the first instance and the second instance at a physical edge of the distributed computing environment, and wherein the service includes a serverless function.

20. A non-transitory computer-readable medium comprising program code that is executable by a processor for causing the processor to:

deploy a first instance of a service in a distributed computing environment using a definition file configured for deploying instances of the service;

receive metrics information describing resource usage by the first instance of the service in the distributed computing environment;

determine a quality-of-service (QoS) constraint for the service; and

subsequent to receiving the metrics information and determining the QoS constraint:

automatically modify the definition file based on the metrics information and the QoS constraint; and

deploy a second instance of the service in the distributed computing environment using the modified definition file, the second instance being configured to more closely satisfy the QoS constraint than the first instance due to the modification to the definition file.