Optimizing a load balancer configuration

• US 10,104,169 B1
• Filed: 12/18/2013
• Issued: 10/16/2018
• Est. Priority Date: 12/18/2013
• Status: Expired due to Fees

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Moniz, Samuel Leonard, Peterson, Kyle Bradley
• Primary Examiner(s): Hoang, Hieu
• Application Number:
  US14/133,529
• Time in Patent Office: 1,763 Days
• Field of Search: 709219, 709225, 709229
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1017 based on a round robin mechanism
• CPC Class Codes: H04L 67/1025 Dynamic adaptation of the criteria on which the server selection is based
• CPC Class Codes: H04L 67/1027 Persistence of sessions during load balancing
• CPC Class Codes: H04L 67/1029 using data related to the state of servers by a load balancer
• CPC Class Codes: H04L 67/61 taking into account QoS or priority requirements

Abstract

Methods, systems, and computer-readable media for optimizing a load balancer configuration are disclosed. A load balancer is configured to perform traffic management for a distributed system. A new component is deployed to the distributed system or a performance deviation is predicted in the distributed system. In response to the initiation of the deployment or the prediction of the performance deviation, the configuration of the load balancer is modified from a first state to a second state. Modifying the configuration of the load balancer modifies the traffic management for the distributed system. If the configuration was changed due to a deployment, the configuration of the load balancer is restored from the second state to the first state after a period of time has elapsed.

Claims

1. A system, comprising:

one or more computing devices configured to implement a load balancer management system, wherein the load balancer management system is configured to;

change a load balancing algorithm of a load balancer from a least connections load balancing algorithm to a round robin load balancing algorithm in response to initiation of a deployment of one or more services or one or more additional hosts to a service-oriented system, wherein the load balancer is configured to direct a plurality of requests to a plurality of hosts in the service-oriented system based on the load balancing algorithm;

after changing the load balancing algorithm of the load balancer from the least connections load balancing algorithm to the round robin load balancing algorithm,

determine, based at least in part on one or more performance metrics for the deployed one or more services or the deployed one or more additional hosts in the service-oriented system, that the deployment of the one or more services or the one or more additional hosts is successful; and

change, in response to the determination that the deployment of the one or more services or the one or more additional hosts is successful, the load balancing algorithm of the load balancer from the round robin load balancing algorithm to the least connections load balancing algorithm for the plurality of hosts in the service-oriented system.

2. The system as recited in claim 1, wherein, in changing the load balancing algorithm of the load balancer from the least connections load balancing algorithm to the round robin load balancing algorithm, the load balancer management system is configured to prioritize availability over latency and cost in the service-oriented system, and wherein, in changing the load balancing algorithm of the load balancer from the round robin load balancing algorithm to the least connections load balancing algorithm, the load balancer management system is configured to prioritize latency and cost over availability in the service-oriented system.

3. The system as recited in claim 1, wherein the load balancer management system is configured to:

collect the one or more performance metrics for the service-oriented system during a period of time in which the load balancing algorithm is the round robin load balancing algorithm.

4. The system as recited in claim 3, wherein a duration of the period of time is determined based on one or more performance metrics.

5. A method, comprising:

initiating deployment of a new component to a distributed system, wherein a load balancer is configured to perform traffic management for the distributed system;

in response to initiating the deployment of the new component, modifying a configuration of the load balancer from a first state to a second state, wherein modifying the configuration of the load balancer modifies the traffic management for the distributed system, wherein modifying the configuration of the load balancer from the first state to the second state comprises changing a load balancing algorithm from a least connections algorithm to a round robin algorithm;

determining, based at least in part on performance metrics for the deployed new component that indicate performance of the deployed new component, that the deployment of the new component is successful; and

restoring, in response to said determining that the deployment of the new component is successful, the configuration of the load balancer from the second state to the first state, wherein restoring the configuration of the load balancer from the second state to the first state comprises changing the load balancing algorithm from the round robin algorithm to the least connections algorithm.

6. The method as recited in claim 5, wherein modifying the configuration of the load balancer from the first state to the second state comprises prioritizing availability over latency and cost in the distributed system, and wherein restoring the configuration of the load balancer to the first state comprises prioritizing latency and cost over availability in the distributed system.

7. The method as recited in claim 5, wherein the new component comprises one or more new services or one or more new versions of one or more services.

8. The method as recited in claim 5, wherein the new component comprises one or more additional hosts.

9. The method as recited in claim 5, further comprising:

collecting the performance metrics for the distributed system after modifying the configuration of the load balancer from the first state to the second state.

10. The method as recited in claim 5, further comprising:

prior to said determining that the deployment of the new component is successful and prior to said restoring the configuration of the load balancer to the first state;

collecting one or more additional performance metrics for the deployed new component;

predicting a performance degradation in the distributed system based on the one or more additional performance metrics;

changing the configuration of the load balancer from the second state to a third state in response to predicting the performance degradation; and

operating the load balancer based on the third state for a period of time.

11. The method as recited in claim 10, wherein said changing the configuration of the load balancer from the second state to a third state comprises changing a respective weight for one or more hosts in a round robin load balancing algorithm.

12. A non-transitory, computer-readable storage medium storing program instructions computer-executable to perform:

collecting one or more performance metrics for one or more hosts of a service-oriented system comprising a plurality of hosts;

predicting a performance degradation in the service-oriented system based on the one or more performance metrics for the one or more hosts;

changing a load balancing configuration of a load balancer from a first load balancing configuration to a second load balancing configuration in response to the predicting the performance degradation based on the one or more performance metrics for the one or more hosts, wherein changing the load balancing configuration of the load balancer from the first load balancing configuration to the second load balancing configuration comprises changing a load balancing configuration from a least connections configuration to a round robin configuration, and wherein the load balancer is configured to direct a plurality of requests to the plurality of hosts in the service-oriented system based on the load balancing configuration;

operating the load balancer based on the second load balancing configuration for a period of time; and

restoring, subsequent to determining that the service-oriented system is free of the performance degradation, the configuration of the load balancer to the first state, wherein restoring the configuration of the load balancer to the first state comprises changing the load balancing algorithm to the least connections algorithm.

13. The non-transitory, computer-readable storage medium as recited in claim 12, wherein the program instructions are further computer-executable to perform:

collecting a plurality of additional performance metrics for the service-oriented system after changing the load balancing configuration from the first load balancing configuration to the second load balancing configuration;

performing an analysis of the plurality of additional performance metrics; and

adjusting the load balancing configuration one or more additional times based on the analysis of the additional performance metrics.

14. The non-transitory, computer-readable storage medium as recited in claim 12, wherein the program instructions are further computer-executable to perform:

collecting one or more additional performance metrics for the service-oriented system during the period of time;

predicting a second performance degradation in the service-oriented system based on the one or more additional performance metrics;

changing the load balancing configuration of the load balancer from the second load balancing configuration to a third load balancing configuration in response to predicting the second performance degradation; and

operating the load balancer based on the third load balancing configuration for a second period of time.

15. The non-transitory, computer-readable storage medium as recited in claim 14, wherein the second load balancing configuration comprises a first set of weights for the plurality of hosts in a round robin load balancing algorithm, and wherein the third load balancing configuration comprises a second set of weights for the plurality of hosts in the round robin load balancing algorithm.

16. The non-transitory, computer-readable storage medium as recited in claim 12, wherein the load balancing configuration comprises a load balancing algorithm, wherein the first load balancing configuration comprises a first load balancing algorithm, and wherein the second load balancing configuration comprises a second load balancing algorithm.

17. The non-transitory, computer-readable storage medium as recited in claim 12, wherein a duration of the period of time is determined based on one or more additional performance metrics.

18. The non-transitory, computer-readable storage medium as recited in claim 12, wherein a duration of the period of time is determined based on user input.

19. The non-transitory, computer-readable storage medium as recited in claim 12, wherein predicting the performance degradation in the service-oriented system comprises determining one or more indicators of an impending performance degradation relative to a threshold.