Cluster computing

• US 10,333,768 B2
• Filed: 02/14/2014
• Issued: 06/25/2019
• Est. Priority Date: 06/13/2006
• Status: Active Grant

• Current Assignee: Advanced Cluster Systems, Inc.
• Original Assignee: Advanced Cluster Systems, Inc.
• Inventors: Tannenbaum, Zvi, Dauger, Dean E.
• Primary Examiner(s): Asres, Hermon
• Application Number:
  US14/181,112
• Publication Number:
  US 20140372586A1
• Time in Patent Office: 1,957 Days
• Field of Search: 709223
• CPC Class Codes: G06F 15/76 Architectures of general purpose stored program computers with program plugboard G06F15/08; multicomputers G06F15/16
• CPC Class Codes: G06F 9/5072 Grid computing
• CPC Class Codes: G06F 9/54 Interprogram communication
• CPC Class Codes: H04L 41/04 Network management architectures or arrangements

Abstract

In some embodiments, a computer cluster system comprises a plurality of nodes and a software package comprising a user interface and a kernel for interpreting program code instructions. In certain embodiments, a cluster node module is configured to communicate with the kernel and other cluster node modules. The cluster node module can accept instructions from the user interface and can interpret at least some of the instructions such that several cluster node modules in communication with one another and with a kernel can act as a computer cluster.

Claims

1. A computer cluster comprising:

a plurality of nodes, wherein each of the plurality of nodes comprises a hardware processor, wherein one or more of the nodes are configured to receive a command to start a cluster initialization process for the computer cluster, and wherein each of the nodes is configured to access a non-transitory computer-readable medium comprising program code for a single-node kernel that, when executed, is capable of causing the hardware processor to evaluate mathematical expressions; and

a mechanism for the nodes to communicate results of mathematical expression evaluation with each other using a peer-to-peer architecture;

wherein the plurality of nodes comprises;

a first node comprising a first hardware processor configured to access a first memory comprising program code for a user interface and program code for a first single-node kernel, the first single-node kernel configured to interpret user instructions and distribute calls to at least one of a plurality of other nodes for execution; and

a second node comprising a second hardware processor with a plurality of processing cores, wherein the second node is configured to receive calls from the first node, execute at least a first mathematical expression evaluation, and communicate a result of the first mathematical expression evaluation to a third node;

wherein the third node comprises a third hardware processor with a plurality of processing cores, wherein the third node is configured to receive the result of the first mathematical expression evaluation from the second node, execute at least a second mathematical expression evaluation using the received result, and communicate the result of the second mathematical expression evaluation to the first node;

wherein the first node is configured to return the result of the second mathematical expression evaluation to the user interface;

wherein one or more of the nodes are configured to;

accept user instructions;

after accepting user instructions, communicate at least some of the user instructions using the mechanism for the nodes to communicate with each other; and

after communicating at least some of the user instructions using the mechanism, communicate at least some of the user instructions to one or more single-node kernels.

2. The computer cluster of claim 1, wherein the single-node kernel comprises a Mathematica kernel.

3. The computer cluster of claim 1, wherein the mechanism comprises a message passing interface.

4. The computer cluster of claim 1, wherein each of the nodes comprises one or more cluster node modules.

5. The computer cluster of claim 1, wherein each single-node kernel is stored in a non-transitory computer-readable medium and configured to accept and execute a request.

6. The computer cluster of claim 5, wherein each of the nodes comprises one or more cluster node modules, wherein each of the cluster node modules comprises instructions stored in a non-transitory computer-readable medium, and wherein the instructions, when executed by the hardware processor, cause the cluster node module to communicate with the single-node kernel and with one or more other cluster node modules.

7. The computer cluster of claim 6, wherein the plurality of cluster node modules act as a cluster.

8. The computer cluster of claim 6, wherein the plurality of cluster node modules communicate with one another to act as a cluster.

9. The computer cluster of claim 8, wherein the computer cluster includes the user interface.

10. The computer cluster of claim 9, wherein each cluster node module accepts instructions from the user interface and interprets one or more of the instructions.

11. The computer cluster of claim 1, wherein the cluster initialization process comprises establishing communication among two or more of the nodes.

12. The computer cluster of claim 1, wherein the cluster initialization process comprises configuring access by one or more of the nodes to a computer-readable medium comprising program code for the single-node kernel.

13. The computer cluster of claim 12, wherein the cluster initialization process comprises establishing message-passing support among the nodes in the cluster.

14. The computer cluster of claim 12, wherein the cluster initialization process comprises launching cluster node modules by the cluster.

15. The computer cluster of claim 14, wherein the cluster initialization process comprises assigning a processor identification number to each of the cluster node modules.

16. The computer cluster of claim 1, wherein the cluster initialization process comprises:

launching cluster node modules by the cluster; and

after launching the cluster node modules, configuring access by one or more of the nodes to a non-transitory computer-readable medium comprising program code for the single-node kernel.

17. The computer cluster of claim 1, wherein the cluster initialization process comprises:

launching cluster node modules by the cluster;

after launching the cluster node modules, establishing communication among two or more of the nodes; and

after establishing communication among two or more of the nodes, assigning a processor identification number to each of the cluster node modules.

18. The computer cluster of claim 1, wherein one or more of the nodes are configured to communicate at least some of the user instructions.

19. The computer cluster of claim 18, wherein one or more of the nodes are configured to communicate at least some of the user instructions to one or more single-node kernels.

20. The computer cluster of claim 1, wherein one or more of the nodes are configured to accept user instructions via one or more of the nodes.

21. The computer cluster of claim 20, wherein one or more of the nodes are configured to communicate at least some of the user instructions using the mechanism for the nodes to communicate with each other.

22. The computer cluster of claim 1, wherein one or more of the nodes are configured to transmit at least some of the user instructions that originate from a user interface.

23. The computer cluster of claim 1, wherein one or more of the nodes are configured to parallelize at least some of the user instructions before communicating at least some of the user instructions to one or more single-node kernels.

24. The computer cluster of claim 1, wherein one or more of the nodes are configured to accept user instructions before communicating at least some of the user instructions to one or more single-node kernels.

25. The computer cluster of claim 1, wherein the plurality of nodes are configured to communicate with one another to interpret and translate commands for execution by a plurality of single-node kernels.

30. The computer cluster of claim 4, wherein each of the plurality of nodes implements asynchronous calls that enable the single-node kernel to perform computation tasks while the cluster node modules are simultaneously communicating with one another.

31. The computer cluster of claim 4, wherein intercommunication among the plurality of single-node kernels during thread execution is enabled by the plurality of cluster node modules, and wherein the computer cluster is configured to permit exchange of information between nodes during the course of a parallel computation.

32. The computer cluster of claim 4, wherein each of the single-node kernels are configured to call a send command involving creating a packet containing:

an expression to be sent as payload; and

where the expression should be sent;

wherein, upon reception of the send command by a local cluster node module associated with the single-node kernel, the local cluster node module is configured to decode the packet and forward a payload to the cluster node module specified in the packet;

wherein each of the single-node kernels is configured to call a receipt command involving creating a packet specifying which message to test for completion and to then wait for a reply expression to evaluate;

wherein, upon reception of the receipt command by the local cluster node module associated with the single-node kernel, the local cluster node module is configured to decode the packet and use a message specifier to search for any matching expressions listed as completed in a received message queue;

wherein, upon determining that such a completed expression is found, the local cluster node module is configured to send the completed expression to a local single-node kernel associated with the cluster node module in response to the receipt command, and

wherein each of the single-node kernels is configured to receive the completed expression and to update variables used by the single-node kernel during evaluation of expressions.

33. The computer cluster of claim 1, wherein the plurality of nodes are configured to permit exchange of information between nodes during the course of parallel computation.

34. The computer cluster of claim 1, wherein each of the plurality of nodes comprises instructions executable by the hardware processor and configured to implement asynchronous behavior, wherein the instructions comprise:

a first instruction to asynchronously send a payload to another node;

a second instruction to asynchronously receive a payload from another node; and

a third instruction to search for a payload matching a message specifier.

26. A computer cluster comprising:

a plurality of nodes, wherein one or more of the nodes are configured to receive;

a command to start a cluster initialization process for the computer cluster, wherein the cluster initialization process comprises establishing communication among two or more of the nodes; and

an instruction from a user interface or a script;

and

a mechanism for the nodes to communicate results of mathematical expression evaluation with each other using asynchronous calls;

wherein each of the nodes is configured to access a non-transitory computer-readable medium comprising program code for a single-node kernel that, when executed, is capable of causing a hardware processor to evaluate mathematical expressions;

wherein the plurality of nodes comprises;

a first node comprising a first hardware processor configured to access a first memory comprising program code for a user interface and program code for a first single-node kernel, the first single-node kernel configured to interpret user instructions and distribute calls to at least one of a plurality of other nodes for execution; and

a second node comprising a second hardware processor with a plurality of processing cores, wherein the second node is configured to receive calls from the first node, execute at least a first mathematical expression evaluation, and communicate a result of mathematical expression evaluation to a third node;

wherein the third node comprises a third hardware processor with a plurality of processing cores, wherein the third node is configured to receive the result of mathematical expression evaluation from the second node, execute at least a second mathematical expression evaluation using the received result, and communicate the result of the second mathematical expression evaluation to the first node;

wherein the first node is configured to return the result of the second mathematical expression evaluation to the user interface or the script;

wherein one or more of the nodes are configured to;

accept user instructions;

after accepting user instructions, communicate at least some of the user instructions using the mechanism for the nodes to communicate with each other; and

after communicating at least some of the user instructions using the mechanism, communicate at least some of the user instructions to one or more single-node kernels.

27. The computer cluster of claim 26, wherein the asynchronous calls comprise a first command to create a first packet containing:

an expression to be sent as payload; and

a target node where the expression should be sent;

wherein the first command is configured to be called from within a single-node kernel;

wherein the single-node kernel is configured to send the first packet to a local cluster node module; and

wherein the local cluster node module is configured to forward the expression to the target node.

28. The computer cluster of claim 27, wherein the asynchronous calls comprise a second command to create a second packet containing:

a location where the expression is expected to be received; and

a sending node from which the expression is expected to be received;

wherein the second command is configured to be called from within a single-node kernel;

wherein the single-node kernel is configured to send the second packet to a local cluster node module; and

wherein the local cluster node module is configured to store the second packet contents in a message receiving queue.

29. A computer cluster comprising:

a plurality of nodes, wherein one or more of the nodes are configured to receive;

a command to start a cluster initialization process for the computer cluster, wherein the cluster initialization process comprises establishing communication among two or more of the nodes; and

an instruction from a user interface or a script;

and

a mechanism for the nodes to communicate results of mathematical expression evaluation with each other;

wherein each of the nodes is configured to access a non-transitory computer-readable medium comprising program code for a single-node kernel that, when executed, is capable of causing a hardware processor to evaluate mathematical expressions;

wherein the plurality of nodes comprises;

a first node comprising a first hardware processor configured to access a first memory comprising program code for a user interface and program code for a first single-node kernel, the first single-node kernel configured to interpret user instructions and distribute calls to at least one of a plurality of other nodes for execution; and

a second node comprising a second hardware processor with a plurality of processing cores, wherein the second node is configured to receive calls from the first node, execute at least a first mathematical expression evaluation, and communicate a result of mathematical expression evaluation to a third node;

wherein the third node comprises a third hardware processor with a plurality of processing cores, wherein the third node is configured to receive the result of mathematical expression evaluation from the second node, execute at least a second mathematical expression evaluation using the received result, and communicate the result of the second mathematical expression evaluation to the first node;

and

wherein the first node is configured to return the result of the second mathematical expression evaluation to the user interface or the script;

wherein one or more of the nodes are configured to;

accept user instructions;

after accepting user instructions, communicate at least some of the user instructions using the mechanism for the nodes to communicate with each other; and

after communicating at least some of the user instructions using the mechanism, communicate at least some of the user instructions to one or more single-node kernels.

35. A computer cluster node for evaluating expressions in parallel with other computer cluster nodes, the computer cluster node comprising:

a hardware processor configured to access one or more non-transitory memory devices comprising program code for a single-node kernel that, when executed, causes the hardware processor to interpret user instructions, to evaluate mathematical expressions, and to produce results of mathematical expression evaluation, wherein the hardware processor comprises multiple processor cores;

a user connection interface configured to receive a command to start a cluster initialization process for a computer cluster;

a mechanism to communicate results of evaluation with other computer cluster nodes using a peer-to-peer architecture; and

program code that, when executed, is capable of causing the hardware processor to;

receive calls from a second node comprising a second hardware processor configured to access a second memory comprising program code for a user interface and program code for a second single-node kernel, the second single-node kernel configured to interpret user instructions and distribute calls to at least one of a plurality of other nodes for execution;

execute, using the hardware processor, at least a first mathematical expression evaluation; and

communicate a result of the first mathematical expression evaluation to a third node comprising a third hardware processor with a plurality of processing cores, wherein the third node is configured to receive the result of mathematical expression evaluation from the computer cluster node, execute at least a second mathematical expression evaluation using the result of the first mathematical expression evaluation, and communicate a result of the second mathematical expression evaluation to the first node;

wherein the user connection interface is configured to return at least one result of mathematical expression evaluation to a user interface or a script; and

wherein the computer cluster node is configured to;

accept user instructions;

after accepting user instructions, communicate at least some of the user instructions using the mechanism for the nodes to communicate with each other; and

after communicating at least some of the user instructions using the mechanism, communicate at least some of the user instructions to the single-node kernel.

36. The computer cluster node of claim 35, wherein the one or more non-transitory memory devices comprise program code for performing a parallel fast Fourier transform, wherein the program code causes the hardware processor to evaluate a command to perform a Fourier transform on an array comprising a first data portion that is stored on the computer cluster node and a second data portion that is not stored on the computer cluster node.

37. The computer cluster node of claim 35, wherein the computer cluster node is configured to permit exchange of information with other computer cluster nodes during the course of parallel computation.

38. The computer cluster node of claim 35, wherein the computer cluster node comprises instructions executable by the hardware processor and configured to implement asynchronous behavior, wherein the instructions comprise:

a first instruction to asynchronously send a payload to another node;

a second instruction to asynchronously receive a payload from another node; and

a third instruction to search for a payload matching a message specifier.

39. The computer cluster node of claim 35, wherein the hardware processor comprises a special purpose microprocessor.