Network monitoring tool for supercomputers

1. A method for monitoring supercomputer nodes, the method comprising:

receiving, by a monitoring device configured to monitor communications between a plurality of nodes of a supercomputer system, node data from a set of one or more nodes allocated to execute a process;

generating, by the monitoring device, a virtual topology representing a physical topology of an architecture of the supercomputer system based on the node data, at least a portion of the node data represents communications between a plurality of nodes of the supercomputer system;

generating, by the monitoring device, a global virtual topology representing one or more virtual topologies based on the node data for communications between each set of one or more nodes of the supercomputer system; and

allocating, by the monitoring device, an available node processor in a second set of one or more nodes to execute a new process according to the global virtual topology.

2. The method of claim 1 , wherein data in the node data is selected from the group consisting of: congestion data, bandwidth data, a process allocation status, an availability status, communications links, inter-node messages, and traffic characteristics.

3. The method of claim 1 , wherein generating the global virtual topology comprises:

for each respective virtual topology, generating, by the monitoring device, the respective virtual topology based on the node data of the respective set of one or more nodes.

4. The method of claim 3 , wherein generating the global virtual topology comprises:

identifying, by the monitoring device, each set of one or more nodes allocated to each process to be executed in the supercomputer system based on the node data, wherein the global virtual topology is based on each virtual topology.

5. The method of claim 1 , wherein generating the global virtual topology comprises:

receiving, by the monitoring device, congestion data for communications between at least two sets of nodes, the global virtual topology based, at least in part, on the congestion data.

6. The method of claim 1 , wherein allocating the new process to the available node:

determining, by the monitoring device, that one or more sets of nodes are allocated to one or more processes based on the node data; and

determining, by the monitoring device, that the available node of the second set of one or more nodes is unallocated to execute the one or more processes based on congestion data.

7. The method of claim 1 , further comprising receiving, by the monitoring device, the node data for each node from one or more switches in the architecture of the supercomputer system.

8. The method of claim 7 , wherein the monitoring device receives the node data from a switch via a switch adapter corresponding to at least node.

9. The method of claim 1 , further comprising generating, by the monitoring device, a graphical display configured to display via a client computer the global topology view and each respective virtual topology.

10. The method of claim 9 , further comprising storing, by the monitoring device, the node data for the set of one or more nodes into a database, wherein the monitoring device generates the graphical display based on the node data stored in the database.

11. A system comprising:

a plurality of supercomputer nodes of a supercomputer system, comprising a plurality of sets of one or more nodes, each set of nodes configured to execute at least one process allocated to the respective set of nodes;

a network monitoring device configured to:

receive node data from a first set of nodes allocated to a process;

generate a virtual topology representing a physical topology of an architecture of the supercomputer system based on the node data, at least a portion of the node data represents communications between the plurality of nodes of the supercomputer system;

generate a global virtual topology representing one or more virtual topologies based on the node data for communications between each set of one or more nodes; and

allocate an available node in a second set of one or more nodes to execute a new process according to the global virtual topology.

12. The system of claim 11 , wherein data in the node data is selected from the group consisting of: congestion data, bandwidth data, a process allocation status, an availability status, communications links, inter-node messages, and traffic characteristics.

13. The system of claim 11 , wherein, to generate the global virtual topology, the network monitoring device is further configured to:

for each respective virtual topology, generate the respective virtual topology based on the node data of the respective set of one or more nodes.

14. The system of claim 13 , wherein, to generate the global virtual topology, the network monitoring device is further configured to:

identify each set of one or more nodes allocated to each process to be executed in the supercomputer system based on the node data, wherein the global virtual topology is based on each virtual topology.

15. The system of claim 11 , wherein, to generate the global virtual topology, the network monitoring device is further configured to:

receive congestion data for communications between at least two sets of nodes, wherein the global virtual topology is based, at least in part, on the congestion data.

16. The system of claim 11 , wherein, to allocate the new process to the available node in the second set of one or more nodes, the network monitoring device is further configured to:

determine that one or more sets of nodes are allocated to one or more processes based on the node data; and

determine that the available node of the second set of one or more nodes is unallocated to execute the one or more processes based on congestion data.

17. The system of claim 11 , wherein the monitoring device is further configured to receive the node data for each node from one or more switches in the architecture of the supercomputer system.

18. The system of claim 17 , wherein the network monitoring device receives the node data from a switch via a switch adapter corresponding to at least node.

19. The system of claim 11 , wherein the network monitoring device is further configured to generate a graphical display configured to display via a client computer the global topology view and each respective virtual topology.

20. The system of claim 19 , wherein the network monitoring device is further configured to store the node data for the set of one or more nodes into a database, wherein the network monitoring device generates the graphical display based on the node data stored in the database.