Reducing energy consumption of servers

• US 20110208875A1
• Filed: 02/24/2010
• Published: 08/25/2011
• Est. Priority Date: 02/24/2010
• Status: Active Grant

• Current Assignee: F5 Networks, Inc. (F5 Incorporated)
• Original Assignee: Crescendo Networks Ltd. (F5 Incorporated)
• Inventors: Hasson, Uri, Shurman, Michael
• Granted Patent: US 8,260,958 B2
• Time in Patent Office: Days
• US Class Current: 709/238
• CPC Class Codes: G06F 1/3203 Power management, i.e. event-based initiation of a power-saving mode
• CPC Class Codes: G06F 1/329 by task scheduling
• CPC Class Codes: G06F 9/505 considering the load
• CPC Class Codes: G06F 9/5094 where the allocation takes into account power or heat criteria power management in computers in general G06F1/3203; thermal management in computers in general G06F1/206
• CPC Class Codes: H04L 67/1023 based on a hash applied to IP addresses or costs
• CPC Class Codes: H04L 67/1025 Dynamic adaptation of the criteria on which the server selection is based
• CPC Class Codes: H04L 67/1029 using data related to the state of servers by a load balancer
• CPC Class Codes: H04L 67/1031 Controlling of the operation of servers by a load balancer, e.g. adding or removing servers that serve requests
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A load balancer, comprising a network interface, a power conservation unit, and a routing module configured to route client requests received through the network interface to a plurality of servers. The power conservation unit is characterized by having a learning mode and a routing mode. In the learning mode one or more operation parameters of the servers are determined for a plurality of different external conditions and for a plurality of different values of one or more operation parameters of the routing module, and to generate a correlation table between the operation parameters of the routing module and the external conditions. In the routing mode the power conservation unit adjusts the operation parameters of the routing module responsive to the external conditions, using the correlation table.

Claims

1. A load balancer, comprising:

a network interface;

a routing module configured to route client requests received through the network interface to a plurality of servers; and

a power conservation unit configured to operate in;

a learning mode in which one or more operation parameters of the servers are determined for a plurality of different external conditions and for a plurality of different values of one or more operation parameters of the routing module, and to generate a correlation table between the operation parameters of the routing module and the external conditions;

and

a routing mode in which the power conservation unit adjusts the operation parameters of the routing module responsive to the external conditions, using the correlation table.

2. A load balancer according to claim 1, wherein the one or more operation parameters of the servers comprise a measure of a response time to client requests.

3. A load balancer according to claim 1, wherein the one or more operation parameters of the routing module comprise the number of servers to which client requests are routed.

4. A load balancer according to claim 1, wherein the one or more operation parameters of the routing module comprises a number of connections used by the routing module to route client requests to the servers.

5. A load balancer according to claim 1, wherein the one or more operation parameters of the routing module comprises a rate at which the routing module forwards client requests to the servers.

6. A load balancer according to claim 1, wherein the routing module is adapted to operate with multi-core servers that self adjust the number of cores they use to their load.

7. A load balancer according to claim 1, wherein during the learning mode the power conservation unit transmits test requests generated by the load balancer to the servers.

8. A load balancer according to claim 1, wherein during the learning mode the power conservation unit selects operation parameters of the routing module corresponding to specific external conditions as the routing module parameters that use the least resources but still provide a response time within a required range.

9. A load balancer according to claim 1, wherein the power conservation unit is adapted to periodically enter the learning mode.

10. A load balancer according to claim 9, wherein the power conservation unit is adapted to enter the learning mode responsive to a determination that the correlation table needs updating.

11. A load balancer according to claim 9, wherein the power conservation unit is adapted to enter the learning mode at predetermined times.

12. A load balancer according to claim 9, wherein during the learning mode the routing module is adapted to route client requests received through the network interface to the servers.

13. A load balancer according to claim 9, wherein during the learning mode the routing module is adapted to generate test requests that are not received through the network interface and to forward the generated test requests to the servers.

14. A load balancer according to claim 13, wherein the routing module is adapted to generate the test requests based on client requests received through the network interface.

15. A method of routing client requests to servers by a load balancer, comprising:

operating the load balancer in a learning mode in which a plurality of load balancer parameter values are tested for each of a plurality of specific external conditions, and accordingly a specific set of load balancer parameter values are selected for each specific external conditions; and

during a routing mode of the load balancer, determining external conditions and selecting load balancer parameter values responsive to the determined external conditions.

16. The method of claim 15, comprising automatically entering the learning mode responsive to a determination that the average response time of the servers is outside a predetermined value range.

17. A method of load balancing, comprising:

receiving client requests by a load balancer;

forwarding the client requests to a plurality of servers adapted to adjust the number of cores they use responsive to their load;

determining a load on the load balancer; and

adjusting the number of connections used to forward the client requests to the servers, responsive to the determined load.

18. The method of claim 17, wherein adjusting the number of connections comprises adjusting to a number selected during a learning mode of the load balancer for the determined load.