Automatic management of low latency computational capacity

• US 9,830,193 B1
• Filed: 09/30/2014
• Issued: 11/28/2017
• Est. Priority Date: 09/30/2014
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Wagner, Timothy Allen, Reque, Sean Philip, Thomas, Dylan Chandler, Manwaring, Derek Steven, Burkett, Bradley Nathaniel
• Primary Examiner(s): Brophy, Matthew
• Application Number:
  US14/502,714
• Time in Patent Office: 1,155 Days
• CPC Class Codes: G06F 2009/45562 Creating, deleting, cloning virtual machine instances
• CPC Class Codes: G06F 2009/45591 Monitoring or debugging support
• CPC Class Codes: G06F 2209/5011 Pool
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects
• CPC Class Codes: G06F 9/5077 Logical partitioning of resources; Management or configuration of virtualized resources specific details on emulation or internal functioning of virtual machines G06F9/455

Abstract

A system for providing automatic management of low latency computational capacity is provided. The system may be configured to maintain a plurality of virtual machine instances. The system may be further configured to identify a trend in incoming code execution requests to execute program code on a virtual compute system, determine, based on the identified trend, that the plurality of virtual machine instances should be adjusted, and adjust the plurality of virtual machine instances based on the identified trend.

Claims

1. A system for providing low-latency computational capacity from a virtual compute fleet, the system comprising:

an electronic data store configured to store at least a program code of a user; and

a virtual compute system comprising one or more hardware computing devices executing specific computer-executable instructions, said virtual compute system in communication with the data store, and configured to at least;

maintain a plurality of virtual machine instances on one or more physical computing devices, wherein the plurality of virtual machine instances comprise;

a warming pool comprising virtual machine instances having one or more software components loaded thereon; and

an active pool comprising virtual machine instances assigned to one or more users,

wherein the virtual compute system is configured to add a first number of virtual machine instances to the warming pool in response to a first virtual machine instance being removed from the warming pool;

monitor incoming code execution requests to execute program codes on the virtual compute system, the incoming code execution requests comprising information usable for identifying the respective program codes to be executed on the virtual compute system;

determine a current rate at which the incoming code execution requests are received;

determine whether the current rate at which the incoming code execution requests are received satisfies a threshold condition for adjusting the warming pool; and

in response to determining that the current rate at which the incoming code execution requests are received satisfies the threshold condition, adjust a rate at which additional virtual machine instances are added to the warming pool such that a second number of virtual machine instances are added to the warming pool in response to a second virtual machine instance being removed from the warming pool, wherein the second number and is different from the first number and determined based on the current rate at which the incoming code execution requests are received.

2. The system of claim 1, wherein the virtual compute system is further configured to:

determine that a utilization level of an under-utilized virtual machine instance in the active pool is below a threshold value, wherein the under-utilized virtual machine instance has one or more containers running one or more program codes therein;

refrain from routing additional code execution requests to the under-utilized virtual machine instance until the one or more containers have completed running the one or more program codes therein; and

terminate the under-utilized virtual machine instance after the one or more containers have completed running the one or more program codes therein.

3. The system of claim 1, wherein the virtual compute system is further configured to create a sub-pool of virtual machine instances within the warming pool, wherein the sub-pool of virtual machine instances are configured to service requests associated with a particular user and the virtual machine instances in the sub-pool have one or more containers have a software configuration that is specific to the particular user.

4. A system, comprising:

a virtual compute system comprising one or more hardware computing devices executing specific computer-executable instructions and configured to at least;

maintain a pool of virtual machine instances on one or more physical computing devices such that a first number of virtual machine instances are added to the plurality pool of virtual machine instances in response to a first virtual machine instance being removed from the pool,

determine a current rate at which incoming code execution requests are received;

determine whether the current rate at which the incoming code execution requests are received satisfies a threshold condition for adjusting the pool of virtual machine instances; and

in response to determining that the current rate at which the incoming code execution requests are received satisfies the threshold condition, adjust a rate at which additional virtual machine instances are added to the pool such that a second number of virtual machine instances are added to the pool in response to a second virtual machine instance being removed from the pool, wherein the second number is different from the first number and determined based on the current rate at which the incoming code execution requests are received.

5. The system of claim 4, wherein the plurality pool of virtual machine instances comprises an active pool of virtual machine instances, and wherein the virtual compute system is further configured to:

determine that a virtual machine instance should be removed from the active pool, wherein the virtual machine instance has one or more containers running one or more program codes therein;

refrain from routing additional code execution requests to the virtual machine instance until the one or more program codes have finished running; and

terminate the virtual machine instance after the one or more program codes have finished running.

6. The system of claim 4, wherein the pool of virtual machine instances comprises a warming pool of virtual machine instances having one or more software components loaded thereon, and wherein the virtual compute system is further configured to:

determine that an available capacity of the warming pool is below a first threshold capacity; and

add virtual machine instances to the warming pool until the available capacity of the warming pool is above the first threshold capacity.

7. The system of claim 6, wherein the virtual compute system is further configured to:

determine that the available capacity of the warming pool is below a second threshold capacity that is lower than the first threshold capacity; and

take at least one additional predetermined action, wherein the at least one additional predetermined action comprises at least one of notifying a system administrator or contacting a service provider providing the virtual machine instances in response to determining that the available capacity of the warming pool is below the second threshold capacity.

8. The system of claim 6, wherein the virtual compute system is further configured to:

forecast a need for additional virtual machine instances based on the current rate; and

adjust a rate at which new virtual machine instances are added to the warming pool.

9. The system of claim 4, wherein the pool of virtual machine instances comprises a warming pool of virtual machine instances having one or more software components loaded thereon, and wherein the virtual compute system is configured to create a sub-pool of virtual machine instances within the warming pool, wherein the sub-pool of virtual machine instances are configured to service code execution requests associated with a particular user.

10. The system of claim 4, wherein the virtual compute system is configured to adjust the pool of virtual machine instances based at least in part on a user policy specified by a user, the user policy indicating an amount of compute capacity that the user desires during a specified time period.

11. The system of claim 4, wherein the virtual compute system is further configured to:

maintain a historical trend associated with a particular user; and

adjust the pool of virtual machine instances based at least in part on the maintained historical trend on the particular user'"'"'s behalf.

12. The system of claim 4, wherein the pool of virtual machine instances comprises a warming pool of virtual machine instances having one or more software components loaded thereon, and wherein the virtual compute system is further configured to:

maintain a first pool of virtual machine instances having a first set of software components loaded thereon and a second pool of virtual machine instances having a second set of software components loaded thereon, wherein the first set of software components include at least one software component not included in the second set of software components;

detect a difference in rates of change in size of the first and second pools; and

adjust a rate at which virtual machine instances are added to at least one of the first pool or the second pool based on the detected difference.

13. A computer-implemented method comprising:

as implemented by one or more computing devices configured with specific executable instructions,

maintaining a pool of virtual machine instances on one or more physical computing devices such that a first number of virtual machine instances are added to the plurality pool of virtual machine instances in response to a first virtual machine instance being removed from the pool;

determining a current rate at which incoming code execution requests are received;

determining whether the current rate at which the incoming code execution requests are received satisfies a threshold condition for adjusting the pool of virtual machine instances; and

in response to determining that the current rate at which the incoming code execution requests are received satisfies the threshold condition, adjusting a rate at which additional virtual machine instances are added to the pool such that a second number of virtual machine instances are added to the pool in response to a second virtual machine instance being removed from the pool, wherein the second number is different from the first number and determined based on the current rate at which the incoming code execution requests are received.

14. The computer-implemented method of claim 13, wherein the pool of virtual machine instances comprises an active pool of virtual machine instances, and wherein the method further comprises:

determining that a virtual machine instance should be removed from the active pool, wherein the virtual machine instance has one or more containers running one or more program codes therein;

refraining from routing additional code execution requests to the virtual machine instance until the one or more program codes have finished running; and

terminating the virtual machine instance after the one or more program codes have finished running.

15. The computer-implemented method of claim 13, wherein the pool of virtual machine instances comprises a warming pool of virtual machine instances having one or more software components loaded thereon, and wherein the method further comprises:

determining that an available capacity of the warming pool is below a first threshold capacity; and

adding virtual machine instances to the warming pool until the available capacity of the warming pool is above the first threshold capacity.

16. The computer-implemented method of claim 15, wherein the method further comprises:

determining that the available capacity of the warming pool is below a second threshold capacity that is lower than the first threshold capacity; and

taking at least one additional predetermined action, wherein the at least one additional predetermined action comprises at least one of notifying a system administrator or contacting a service provider providing the virtual machine instances in response to determining that the available capacity of the warming pool is below the second threshold capacity.

17. The computer-implemented method of claim 13, wherein the pool of virtual machine instances comprises a warming pool of virtual machine instances having one or more software components loaded thereon, and wherein the method further comprises creating a sub-pool of virtual machine instances within the warming pool, wherein the sub-pool of virtual machine instances are configured to service code execution requests associated with a particular user.

18. Non-transitory physical computer storage storing computer executable instructions that, when executed by one or more computing devices, configure the one or more computing devices to:

maintain a pool of virtual machine instances on one or more physical computing devices such that a first number of virtual machine instances are added to the pool of virtual machine instances in response to a first virtual machine instance being removed from the pool;

determine a current rate at which incoming code execution requests are received;

determine whether the current rate at which the incoming code execution requests are received satisfies a threshold condition for adjusting the plurality pool of virtual machine instances; and

in response to determining that the current rate at which the incoming code execution requests are received satisfies the threshold condition, adjust a rate at which additional virtual machine instances are added to the pool such that a second number of virtual machine instances are added to the pool in response to a second virtual machine instance being removed from the pool, wherein the second number is different from the first number and determined based on the current rate at which the incoming code execution requests are received.

19. The non-transitory physical computer storage of claim 18, wherein the pool of virtual machine instances comprises an active pool of virtual machine instances, and wherein the instructions further configure the one or more computing devices to:

determine that a virtual machine instance should be removed from the active pool, wherein the virtual machine instance has one or more containers running one or more program codes therein;

refrain from routing additional code execution requests to the virtual machine instance until the one or more program codes have finished running; and

terminate the virtual machine instance after the one or more program codes have finished running.

20. The non-transitory physical computer storage of claim 18, wherein the pool of virtual machine instances comprises a warming pool of virtual machine instances having one or more software components loaded thereon, and wherein the instructions further configure the one or more computing devices to:

determine that an available capacity of the warming pool is below a first threshold capacity; and

add virtual machine instances to the warming pool until the available capacity of the warming pool is above the first threshold capacity.

21. The non-transitory physical computer storage of claim 20, wherein the instructions further configure the one or more computing devices to:

determine that the available capacity of the warming pool is below a second threshold capacity that is lower than the first threshold capacity; and

take at least one additional predetermined action, wherein the at least one additional predetermined action comprises at least one of notifying a system administrator or contacting a service provider providing the virtual machine instances in response to determining that the available capacity of the warming pool is below the second threshold capacity.

22. The non-transitory physical computer storage of claim 18, wherein the pool of virtual machine instances comprises a warming pool of virtual machine instances having one or more software components loaded thereon, and wherein the instructions further configure the one or more computing devices to create a sub-pool of virtual machine instances within the warming pool, wherein the sub-pool of virtual machine instances are configured to service code execution requests associated with a particular user.