System and methods for load placement in data centers

1. A system for operating a plurality of switches forming a network, the system comprising:

an open flow controller coupled to the switches;

wherein the open flow controller is configured to:

receive a notification of a start of a flow;

determine traffic rates on a first link connecting a first switch and a second switch in the network; and

in response to determining that the traffic rates on the first link exceed a threshold:

send a command to a forwarding top of rack switch to perform a push label operation on first network traffic in the flow so that the first network traffic is re-routed away from the first link;

send a command to the first and second switches to perform a swap label operation on second network traffic to re-route the second network traffic away from the first link; and

send a command to a receiving top of rack switch to perform a pop label operation on the first network traffic.

2. The system of claim 1 wherein the switches are configured to perform bidirectional forwarding detection (BFD) with a neighboring node in a data center network by exchanging one or more handshake packets with the neighboring node.

3. The system of claim 1 wherein each of the switches comprises an open flow agent, the open flow agent configured to receive respective commands from the open flow controller.

4. The system of claim 3 wherein the open flow agent comprises:

a processor circuit to compute a fail re-route (FRR) path set; and

a memory circuit to store the FRR path set;

wherein the FRR path set provides one or more link-specific backup routes in the event of a link failure without involving the open flow controller at the time of the link failure.

5. The system of claim 4 wherein: the FRR path set includes an address for the switch associated to the open flow agent.

6. A computer program product comprising a non-transitory computer readable medium having computer readable and executable code for instructing a processor in a management unit for a plurality of switches forming a network to perform a method, the method comprising:

performing a discovery of a network topology of the network;

computing a label switch path between a forwarding switch and a receiving switch in the network;

computing a fail re-route (FRR) path set for a first one of the switches based on the discovery of the network topology, wherein the FRR path set includes a link-specific FRR path for each link in the label switch path between the first one of the switches and the receiving switch; and

providing the FRR path set to the first one of the switches for use by the first one of the switches to provide one or more link-specific backup routes in the event of a link failure without involving the management unit at the time of the link failure;

wherein computing the label switch path comprises sending a command to perform a push label operation to the forwarding switch.

7. The computer program product of claim 6 wherein computing the label switch path further comprises: sending a command to perform a swap label operation to one or more switches in an aggregation layer of the network.

8. The computer program product of claim 6 wherein computing the label switch path further comprises: sending a command to perform a pop label operation to the receiving switch.

9. The computer program product of claim 6 further comprising:

allocating a label space for each switch in the network; and

turning a label forwarding operation ‘ON’ in each of the switches.

10. The computer program product of claim 6 further comprising maintaining a label space in a memory circuit in a switch from the network.

11. The computer program product of claim 6 further comprising:

monitoring traffic loads in the network; and

rerouting traffic around overloaded links based on the monitored traffic loads.

12. The computer program product of claim 6 further comprising computing the label switch path when a traffic flow starts.

13. The computer program product of claim 12 further comprising releasing the label switch path when the traffic flow stops.

14. A network managing device comprising:

a link to a plurality of switches;

a processor circuit configured to discover a topology of the network, and to compute a label switch path between a forwarding switch and a receiving switch in the network; and

a memory circuit to store the label switch path and the discovered topology of the network;

wherein the processor circuit is further configured to:

compute a fail re-route (FRR) path set for a first switch based on the discovered topology, wherein the FRR path set includes a link-specific FRR path for each link in the label switch path between the first switch and the receiving switch; and

provide the FRR path set to the first switch for use by the first switch to provide one or more link-specific backup routes in the event of a link failure without involving the network managing device at the time of the link failure;

wherein to compute the label switch path the processor circuit is further configured to send a command to perform a push label operation to the forwarding switch.

15. The network managing device of claim 14 wherein the processor circuit is further configured to provide the label switch path to each of the plurality of switches along the label switch path through an agent circuit in each of the plurality of switches along the label switch path.

16. The network managing device of claim 14 wherein to compute a label switch path the processor circuit is configured to:

receive a failure report from the first switch, the failure report reporting the link failure.

17. The network managing device of claim 14 wherein to compute the label switch path the processor circuit is further configured to send a command to perform a swap label operation to one or more switches on the label switch path.

18. The network managing device of claim 14 wherein to compute the label switch path the processor circuit is further configured to send a command to perform a pop label operation to the receiving switch.