IMPLEMENTING REDUNDANCY ON INFINIBAND (IB) NETWORKS

• US 20140334290A1
• Filed: 07/24/2014
• Published: 11/13/2014
• Est. Priority Date: 08/30/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: GAVRILOV, Constantine, MACHULSKY, Zorik, SHALEV, Leah
• Granted Patent: US 9,369,334 B2
• Time in Patent Office: Days
• US Class Current: 370/225
• CPC Class Codes: H04L 2101/622 Layer-2 addresses, e.g. medium access control [MAC] addresses
• CPC Class Codes: H04L 41/0654 using network fault recovery ring fault isolation or reconfiguration in loop networks without recovery actions by a network management system H04L12/437
• CPC Class Codes: H04L 41/0668 by dynamic selection of recovery network elements, e.g. replacement by the most appropriate element after failure
• CPC Class Codes: H04L 45/28 using route fault recovery
• CPC Class Codes: H04L 61/103 across network layers, e.g. resolution of network layer into physical layer addresses or address resolution protocol [ARP]
• CPC Class Codes: H04L 61/2514 between local and global IP addresses

Abstract

Methods, systems, and physical computer-readable storage medium are provided to improve efficiency of systems operating on internet protocol (IP) over Infiniband (IB) networks. In an embodiment, by way of example only, a method is provided for implementing redundancy for IB networks. The method includes detecting a failure at a first source port of a source node having a plurality of source ports, selecting a new source port, sending an address resolution protocol (ARP) message to one or more other nodes in communication with the source node including an IP address of the source node and IB address of the new source port of the source node, and re-balancing host traffic.

Claims

1. A method of implementing redundancy for Infiniband (IB) networks, the method comprising:

detecting a failure at a first source port of a source node having a plurality of source ports;

selecting a new source port of the source node;

sending an address resolution protocol (ARP) announce message to one or more other nodes in communication with the source node, the ARP announce message including an IP address of the second source node and an IB address of the new source port of the source node; and

re-balancing host traffic.

2. The method of claim 1, further comprising identifying from a plurality of source ports of the source node the new source port of the source node with which to receive traffic from the one or more other nodes, after the step of detecting.

3. The method of claim 1, the step of sending the ARP announce message comprises identifying the one or more other nodes in communication with the source node from a global ARP cache.

4. The method of claim 3, wherein the global ARP cache comprises a first level and a second level, the first level includes a plurality of IP addresses and corresponding IP over IB GUIDs of at least one of the one or more other nodes, and the second level includes the IP over IB GUIDs and corresponding IB addresses of at least one of the one or more other nodes.

5. The method of claim 4, further comprising updating the global ARP cache to invalidate references to an address handle associated with the first source port in all entries in the global ARP cache.

6. The method of claim 1, wherein the step of rebalancing comprises identifying one or more nodes that communicate with the source node and notifying the one or more nodes of the address of the new source port.

7. The method of claim 6, wherein the step of notifying comprises multicasting the IB address of the new source port.

8. The method of claim 1, wherein:

the step of rebalancing comprises identifying one or more nodes that communicate with the source node and unicasting the IB address of the new source port to a destination node; and

the step of sending the ARP announce message comprises identifying the one or more nodes in communication with the source node from a global ARP cache.

9. An Infiniband system comprising:

a plurality of nodes, each node including more than one port;

an IB switch connected to the plurality of nodes; and

a processor configured

to detect a failure at a first port of a first node;

to select a new source port of the first node;

to send an address resolution protocol (ARP) announce message to one or more nodes of the plurality of IB nodes in communication with the first node, the ARP announce message including an IP address of the first node and IB address of the new port of the first node; and

to re-balance host traffic across the plurality of IB nodes.

10. The Infiniband system of claim 9, further comprising a plurality of IB switches connecting at least two nodes.

11. The Infiniband system of claim 9, wherein each node includes a two-level ARP cache, the two level ARP cache has a first level and a second level, the first level includes a plurality of IP addresses and corresponding IP over IB GUIDs of at least one of the one or more other nodes, and the second level includes the IP over IB GUIDs and corresponding IB addresses of at least one of the one or more other nodes.

12. A physical computer-readable storage medium comprising a computer program product for implementing redundancy for Infiniband (IB) networks, the physical computer-readable storage medium comprising:

computer code for detecting a failure at a first source port of a source node having a plurality of source ports;

computer code for selecting a new source port;

computer code for sending an address resolution protocol (ARP) announce message to one or more other nodes in communication with the source node, the ARP announce message including an IP address of the source node and IB address of the new source port of the source node; and

computer code for re-balancing host traffic.

13. The physical computer-readable storage medium of claim 12, further comprising computer code for identifying from a plurality of source ports of the source node the new source port of the source node with which to receive traffic from the one or more other nodes.

14. The physical computer-readable storage medium of claim 12, further comprising computer code for identifying the one or more other nodes in communication with the source node from a global ARP cache.

15. The physical computer-readable storage medium of claim 14, wherein the global ARP cache is configured to include a first level and a second level, the first level includes a plurality of IP addresses and corresponding IP over IB GUIDs of at least one of the one or more other nodes, and the second level includes the IP over IB GUIDs and corresponding IB addresses of at least one of the one or more other nodes.

16. The physical computer-readable storage medium of claim 15, further comprising computer code for updating the global ARP cache to invalidate references to an address handle associated with the first source port in all entries in the global ARP cache.

17. The physical computer-readable storage medium of claim 12, further comprising computer code for identifying one or more nodes that communicate with the source node and notifying the one or more nodes of the address of the new source port.

18. The physical computer-readable storage medium of claim 17, further comprising computer code for multicasting the address of the new source port.

19. The physical computer-readable storage medium of claim 12, wherein:

the computer code for rebalancing comprises computer code for identifying one or more nodes that communicate with the source node and unicasting the address of the new source port to a destination node; and

the computer code for sending the ARP announce message comprises computer code for identifying the one or more nodes in communication with the source node from a global ARP cache.