Systems and methods to save power in data-center networks
View Patent ↗A network may assign or reassign data paths to allow the network to shut down servers or switches and meet network demand at a lower energy cost. By assigning or reassigning data paths, network demand is met using only a subset of the network elements, allowing a subset of servers and switches to be temporarily shut down, thereby saving power while maintaining the ability to meet network demand.
1. A computer-implemented method to reduce power consumption by a computer network, comprising:
determining a set of minimum network elements to carry network traffic;
identifying a logical topology for the set of minimum network elements;
reconfiguring an optical circuit switch to implement the logical topology based on the set of minimum network elements;
applying a constraint-based routing model to determine how to reroute traffic over the logical topology for the set of minimum network elements from a high-power-cost network element to a low-power-cost network element, the constraints including carrying network traffic and reducing power consumption;
rerouting network traffic over the logical topology from the high-power-cost network element to the low-power-cost network element, based on the constraint-based routing model;
responsive to the rerouting, identifying and placing a network element bearing reduced network traffic into a low-power state;
monitoring network element utilization to determine whether more network elements are required to be activated to carry network traffic; and
identifying and activating a low-power-cost network element to remove the low-power-cost network element from a low-power state and routing traffic to the activated low-power-cost network element.
2. The method of claim 1 , wherein the computer network is a data-center network.
3. The method of claim 1 , wherein the computer network is controlled by an OpenFlow controller.
4. The method of claim 1 , wherein a network element may be a server, link, or switch.
5. The method of claim 4 , further comprising associating each network element with a power cost.
6. The method of claim 5 , further comprising calculating the power cost of each network element by summing a power consumption of the network element with a ratio of a power consumption of a next-tier network element connected to the network element to a number of links active on the next-tier network element.
7. The method of claim 1 , wherein the low-power state deactivates the network element.
8. The method of claim 1 , wherein the low-power state deactivates a plurality of ports of the network element.
9. The method of claim 8 , wherein the low-power state deactivates a component associated with the plurality of ports.
10. The method of claim 9 , wherein the component may be a line card, an application-specific integrated circuit, or a network processor.
11. The method of claim 1 , further comprising activating the low-power-cost network element using a Wake-On-LAN-enabled network card controller.
12. A computer-implemented method to reduce power consumption by a computer network, comprising:
providing a power cost associated with a network element;
determining a set of minimum network elements to carry network traffic;
identifying a logical topology for the minimum set of network elements;
reconfiguring an optical circuit switch to implement the logical topology based on the set of minimum network elements;
rerouting network traffic over the logical topology from a high-power-cost network element to a low-power-cost network element over the logical topology for the set of minimum network elements according to a linear programming model of constraint-based routing, the constraints including carrying network traffic and reducing power consumption;
responsive to the rerouting, identifying and placing a network element bearing reduced network traffic into a low-power state;
monitoring network element utilization to determine whether more network elements are required to carry network traffic; and
identifying and activating a low-power-cost network element to remove the low-power-cost network element from a low-power state and routing traffic to the activated low-power cost network element.
13. The method of claim 12 , wherein the computer network is a data-center network.
14. The method of claim 12 , wherein the computer network is controlled by an OpenFlow controller.
15. The method of claim 12 , wherein a network element may be a server, link, or switch.
16. The method of claim 15 , wherein a power cost of a network element is the sum of a power consumption of the network element and a ratio of a power consumption of a next-tier network element connected to the network element to a number of links active on the next-tier network element.
17. The method of claim 12 , wherein the low-power state deactivates the network element.
18. The method of claim 12 , wherein the low-power state deactivates a plurality of ports of the network element.
19. The method of claim 18 , wherein the low-power state deactivates a component associated with the plurality of ports.
20. The method of claim 19 , wherein the component may be a line card, an application-specific integrated circuit, or a network processor.
21. The method of claim 12 , further comprising activating the low-power-cost network element using a Wake-On-LAN-enabled network card controller.
22. A system to reduce power consumption by a computer network having an optical circuit switch and a software-configured packet switch, comprising:
an optical circuit switch manager for programming the optical circuit switch to implement a logical topology for the computer network;
a network element manager for monitoring a status of a network element and for programming the software-configured packet switch with a traffic routing plan;
a remote-access power switch for changing a power state of the network element;
a power manager for monitoring a power consumption of the network element and changing the power state of the network element; and
a solver for identifying an alternative logical topology for a set of minimum network elements based on the status of the network element and a network performance target, for generating one or more alterations to the logical topology, and for applying a constraint-based routing model to determine how to reroute traffic over a set of minimum network elements from a high-power-cost network element to a low-power-cost network element, the constraints including carrying network traffic and reducing power consumption the traffic routing plan.
23. The system of claim 22 , wherein the computer network is a data-center network.
24. The system of claim 22 , wherein a network element may be a server, link, or switch.
25. The system of claim 22 , wherein the network element manager is an OpenFlow controller. The method of claim 12 , further comprising activating the low-power-cost network element using a Wake-On-LAN-enabled network card controller.
26. The system of claim 22 , wherein the remote-access power switch is further configured for deactivating the network element.
27. The system of claim 22 , wherein the remote-access power switch is further configured for deactivating a plurality of ports of the network element.
28. The system of claim 27 , wherein the remote-access power switch is further configured for deactivating a component associated with the plurality of ports.
29. The method of claim 28 , wherein the component may be a line card, an application-specific integrated circuit, or a network processor.
30. The system of claim 22 , wherein the remote-access power switch is a Wake-On-LAN-enabled network card controller.