System and method for efficient network reconfiguration in fat-trees

1. A method for supporting efficient reconfiguration of an interconnection network having a pre-existing routing comprising:

providing, at one or more computers, including one or more microprocessors,

a plurality of switches, the plurality switches comprising one or more leaf switches, wherein each of the one or more leaf switches comprise a plurality of ports, and

a plurality of end nodes, wherein the plurality of end nodes are interconnected via the one or more leaf switches;

detecting, by a subnet manager, a reconfiguration triggering event, the reconfiguration triggering event comprising at least one of a changed network traffic pattern and a drop in Quality-of-Service (QOS);

computing, by the subnet manager, a new routing for the interconnection network, wherein the computing by the subnet manager of the new routing for the interconnection network takes into consideration the pre-existing routing and selects the new routing for the interconnection network that is closest to the pre-existing routing; and

reconfiguring the interconnection network according to the new routing for the interconnection network;

wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving the pre-existing port selections for the plurality of end nodes, so long as said preserving the pre-existing port selections for the plurality of end nodes does not conflict with QOS requirements and so long as said preserving the pre-existing port selections for the plurality of end nodes does not conflict current traffic patterns.

2. The method of claim 1 , wherein the new routing for the interconnection network that is closest to the pre-existing routing comprises a route wherein a fewest number of linear forwarding table updates are performed.

3. The method of claim 2 , wherein the reconfiguration triggering event additionally comprises a topology change of the interconnection network.

4. The method of claim 3 , wherein the topology change of the interconnection network comprises at least one of node failure and link failure.

5. The method of claim 4 , wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving pre-existing port selections for the plurality of end nodes, so long as said preserving pre-existing port selections for the plurality of end nodes maintains balance between routes on available port groups.

6. The method of claim 5 , wherein the computing by the subnet manager of the new routing for the interconnection network is deterministic and calculates the new routing backwards, starting at a destination node of the plurality of end nodes.

7. The method of claim 5 , wherein the balance between routes on available ports is maintained by using maximum counters.

8. The method of claim 1 , wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving pre-existing port selections for the plurality of end nodes, so long as said preserving pre-existing port selections for the plurality of end nodes maintains balance between routes on available port groups.

9. The method of claim 8 , wherein the computing by the subnet manager of the new routing for the interconnection network is deterministic and calculates the new routing backwards, starting at a destination node of the plurality of end nodes.

10. The method of claim 9 , wherein the balance between routes on available ports is maintained by using maximum counters.

11. The method of claim 1 , wherein the interconnection network comprises an InfiniBand subnet.

12. A system for supporting efficient reconfiguration of an interconnection network having a pre-existing routing comprising:

one or more microprocessors;

one or more computers;

a plurality of switches, the plurality switches comprising at least one leaf switch, wherein each of the one or more switches comprise a plurality of ports;

a plurality of end nodes, wherein the plurality of end nodes are interconnected via the one or more switches; and

a subnet manager, wherein the subnet manager is configured to detect a reconfiguration triggering event, the reconfiguration triggering event comprising at least one of a changed network traffic pattern and a drop in Quality-of-Service (QOS);

wherein the subnet manager computes a new routing for the interconnection network, wherein the computing by the subnet manager of the new routing for the interconnection network takes into consideration the pre-existing routing and selects the new routing for the interconnection network that is closest to the pre-existing routing; and

wherein the interconnection network is reconfigured according to the new routing for the interconnection network;

wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving the pre-existing port selections for the plurality of end nodes, so long as said preserving the pre-existing port selections for the plurality of end nodes does not conflict with QOS requirements and so long as said preserving the pre-existing port selections for the plurality of end nodes does not conflict current traffic patterns.

13. The system of claim 12 , wherein the new routing for the interconnection network that is closest to the pre-existing routing comprises a route wherein a fewest number of linear forwarding table updates are performed.

14. The system of claim 13 , wherein the reconfiguration triggering event additionally comprises a topology change of the interconnection network.

15. The system of claim 14 , wherein the topology change of the interconnection network comprises at least one of node failure and link failure.

16. The system of claim 15 , wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving pre-existing port selections for the plurality of end nodes, so long as said preserving pre-existing port selections for the plurality of end nodes maintains balance between routes on available port groups.

17. The system of claim 16 , wherein the computing by the subnet manager of the new routing for the interconnection network is deterministic and calculates the new routing backwards, starting at a destination node of the plurality of end nodes.

18. The system of claim 16 , wherein the balance between routes on available ports is maintained by using maximum counters.

19. The system of claim 12 , wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving pre-existing port selections for the plurality of end nodes, so long as said preserving pre-existing port selections for the plurality of end nodes maintains balance between routes on available port groups.

20. The system of claim 19 , wherein the computing by the subnet manager of the new routing for the interconnection network is deterministic and calculates the new routing backwards, starting at a destination node of the plurality of end nodes.

21. The system of claim 20 , wherein the balance between routes on available ports is maintained by using maximum counters.

22. The system of claim 12 , wherein the interconnection network comprises an InfiniBand subnet.

23. A non-transitory computer readable storage medium, including instructions stored thereon for supporting efficient reconfiguration of an interconnection network having a pre-existing routing, which when read and executed by one or more computers cause the one or more computers to perform steps comprising:

providing, at one or more computers, including one or more microprocessors,

a plurality of switches, the plurality switches comprising one or more leaf switches, wherein each of the one or more leaf switches comprise a plurality of ports, and

a plurality of end nodes, wherein the plurality of end nodes are interconnected via the one or more leaf switches;

detecting, by a subnet manager, a reconfiguration triggering event, the reconfiguration triggering event comprising at least one of a changed network traffic pattern and a drop in Quality-of-Service (QOS);

computing, by the subnet manager, a new routing for the interconnection network, wherein the computing by the subnet manager of the new routing for the interconnection network takes into consideration the pre-existing routing and selects the new routing for the interconnection network that is closest to the pre-existing routing; and

reconfiguring the interconnection network according to the new routing for the interconnection network;

wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving the pre-existing port selections for the plurality of end nodes, so long as said preserving the pre-existing port selections for the plurality of end nodes does not conflict with QOS requirements and so long as said preserving the pre-existing port selections for the plurality of end nodes does not conflict current traffic patterns.

24. The non-transitory computer readable storage medium of claim 23 , wherein the new routing for the interconnection network that is closest to the pre-existing routing comprises a route wherein a fewest number of linear forwarding table updates are performed.

25. The non-transitory computer readable storage medium of claim 24 , wherein the reconfiguration triggering event additionally comprises a topology change of the interconnection network.

26. The non-transitory computer readable storage medium of claim 25 , wherein the topology change of the interconnection network comprises at least one of node failure and link failure.

27. The non-transitory computer readable storage medium of claim 26 , wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving pre-existing port selections for the plurality of end nodes, so long as said preserving pre-existing port selections for the plurality of end nodes maintains balance between routes on available port groups.

28. The non-transitory computer readable storage medium of claim 27 , wherein the computing by the subnet manager of the new routing for the interconnection network is deterministic and calculates the new routing backwards, starting at a destination node of the plurality of end nodes.

29. The non-transitory computer readable storage medium of claim 27 , wherein the balance between routes on available ports is maintained by using maximum counters.

30. The non-transitory computer readable storage medium of claim 23 , wherein the computing by the subnet manager of the new routing for the interconnection network comprises:

preserving pre-existing port selections for the plurality of end nodes, so long as said preserving pre-existing port selections for the plurality of end nodes maintains balance between routes on available port groups.

31. The non-transitory computer readable storage medium of claim 30 , wherein the computing by the subnet manager of the new routing for the interconnection network is deterministic and calculates the new routing backwards, starting at a destination node of the plurality of end nodes.

32. The non-transitory computer readable storage medium of claim 31 , wherein the balance between routes on available ports is maintained by using maximum counters.

33. The non-transitory computer readable storage medium of claim 23 , wherein the interconnection network comprises an InfiniBand subnet.