TRANSACTIONAL GRAPH-BASED COMPUTATION WITH ERROR HANDLING

• US 20090030863A1
• Filed: 07/25/2008
• Published: 01/29/2009
• Est. Priority Date: 07/26/2007
• Status: Active Grant

• Current Assignee: Ab Initio Technology LLC (Ab Initio Software Corporation)
• Original Assignee: Ab Initio Software Corporation
• Inventors: Wholey, Joseph Skeffington III, Stanfill, Craig W.
• Granted Patent: US 8,706,667 B2
• Time in Patent Office: Days
• US Class Current: 706/45
• CPC Class Codes: G06F 9/4494 data driven
• CPC Class Codes: G06F 9/466 Transaction processing
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

Processing transactions using graph-based computations includes determining that at least one of a plurality of graph elements of a computation graph of a set of one or more computation graphs includes a computation to be performed for a given transaction, associating the given transaction with an instance of the computation graph that includes reusable computation elements associated with respective graph elements, and executing the graph to perform the computation.

Claims

1. A method for processing transactions using graph-based computations, the method including:

determining that at least one of a plurality of graph elements of a computation graph of a set of one or more computation graphs includes a computation to be performed for a given transaction,

associating the given transaction with an instance of the computation graph that includes reusable computation elements associated with respective graph elements, and

executing the graph to perform the computation.

2. The method of claim 1 wherein at least some instances of the graphs in the set of computation graphs share one or more of the computation elements.

3. The method of claim 1 wherein the computation elements include computations executed by at least one of an operating system process and a process thread.

4. The method of claim 1 wherein the graph elements include vertices of the computation graphs.

5. The method of claim 1 wherein associating the transaction with an instance of the computation graph includes assigning a computation element corresponding to each graph element in the computation graph to the instance of the computation graph before beginning executing the graph elements.

6. The method of claim 1 wherein associating the transaction with an instance of the computation graph includes assigning a computation element corresponding to a graph element in the computation graph to the instance of the computation graph after executing another graph element using a computation element already assigned to the instance.

7. The method of claim 1 in which at least two of the graph elements use a common resource, and executing the graph to perform the computation includes assigning each of the graph elements using the common resource to a single computation element.

8. The method of claim 7 in which the single computation element is already initiated when the graph elements are assigned to the computation element.

9. The method of claim 7 in which the common resource includes a database.

10. The method of claim 7 in which the common resource includes a specific port.

11. The method of claim 1 wherein processing the transaction includes receiving a request for the transaction.

12. The method of claim 1 also including

determining that the same computation graph is associated with a computation to be performed for a second transaction,

associating the second transaction with a second instance of the computation graph, and

executing the second instance of the graph to perform the computation for the second transaction.

13. The method of claim 12 wherein the computations for transactions performed using different instances of computation graphs are performed in a time-interleaved manner.

14. The method of claim 12 wherein multiple transactions are processed concurrently.

15. The method of claim 12 wherein each transaction is associated with one or more work elements that are processed according to the corresponding computation graph.

16. The method of claim 15 wherein at least some transactions are each associated with one work element that is processed according to the corresponding computation graph.

17. The method of claim 1 further including forming multiple instances of at least some of the computation graphs.

18. The method of claim 1 further including identifying that an error has occurred in the performing of a computation for one of the transactions, and continuing the performing of a computation for another one of the transactions.

19. The method of claim 1 wherein the processing of a first transaction of the plurality of transactions starts at a first time, and the processing of a second transaction of the plurality of transactions starts at a second time later than the first time, the method further including completing the performing of the computation for the second transaction before completing the performing of the computation for the first transaction.

20. A system for processing transactions using graph-based computations, the system including:

means for determining that at least one of a plurality of graph elements of a computation graph of a set of one or more computation graphs includes a computation to be performed for a transaction,

means for associating the given transaction with an instance of the computation graph that includes reusable computation elements associated with respective graph elements, and

means for executing the graph to perform the computation.

21. A computer-readable medium storing a computer program for processing transactions using graph-based computations, the computer program including instructions for causing a computer system to:

determine that at least one of a plurality of graph elements of a computation graph of a set of one or more computation graphs includes a computation to be performed for a given transaction,

associate the given transaction with an instance of the computation graph that includes reusable computation elements associated with respective graph elements, and

execute the graph to perform the computation.

22. A method for processing graph-based computations including:

within a graph including vertices representing graph components that process work elements according to links joining the vertices,

providing at least one error-handling graph component configured to provide error information to a process external to the graph, and

processing data, including, in response to a graph component encountering an error while processing, redirecting processing to the error-handling graph component including directing at least some of the work elements to the error-handling component according to at least one link to a vertex representing the error-handling component.

23. The method of claim 22 in which redirecting processing to the error-handling graph component includes removing work elements from at least one input queue.

24. The method of claim 22 in which redirecting processing to the error-handling graph component includes processing the work elements directed to the error-handling graph component.

25. The method of claim 24 in which processing the work elements directed to the error-handling graph component includes rolling back changes to a database made prior to the error.

26. The method of claim 24 in which processing the data includes, for graph components not included in handling the error, discarding work elements directed to those graph components.

27. The method of claim 22 also including providing a sub-graph, the sub-graph including an error-handling sub-graph component configured to provide an error code as an output of the sub-graph.

28. The method of claim 25 in which if output provided by the sub-graph indicates that an error occurred in the sub graph, redirecting processing to the error-handling graph component.

29. The method of claim 22 in which redirecting processing to the error-handling graph component includes communicating, from the graph component that encountered the error, to the error-handling graph component, work elements that the graph component was processing when the error occurred.

30. The method of claim 29 in which the work elements are communicated according to the link to the vertex representing the error-handling component.

31. The method of claim 29 in which redirecting processing to the error-handling graph component includes communicating, from the graph component that encountered the error, to the error-handling graph component, reporting information about the error.

32. The method of claim 31 in which the reporting information is communicated according to an implicit connection between the graph component that encountered the error and the error-handling component.

33. The method of claim 32 also including revealing the implicit connection as an explicit link between a vertex representing the graph component that encountered the error and a vertex representing the error-handling component in response to a user request.

34. The method of claim 22 in which providing the error-handling graph component includes providing a plurality of error-handling graph components, and redirecting processing to the error-handling graph component includes selecting an error-handling graph component based on output provided from the graph component that encountered the error.

35. The method of claim 22 in which processing the data also includes, if a graph component encounters an error while processing, outputting an identification of a work element that caused the error.

36. The method of claim 22 in which processing includes:

enabling a first component of the graph;

disabling the error-handling component; and

for each component downstream of the first component other than the error-handling component, enabling the component if a component immediately upstream of the component is enabled.

37. The method of claim 36 in which redirecting processing to the error-handling graph component includes:

stopping execution of each enabled graph component,

disabling the component that encountered the error;

enabling the error-handling component;

disabling components downstream of the component that encountered the error that are not downstream of the error-handling component; and

enabling components upstream of the error-handling component.

38. The method of claim 22 in which redirecting processing to the error-handling graph component includes, where the error occurred in a first component,

if the error occurs under a first condition, directing process flow from the first component to a first error-handling component upstream of the first component,

and if the error occurs under a second condition, directing process flow from the first component to a second error-handling component downstream of the first component.

39. The method of claim 38 in which the first condition is that a counter is below a limit.

40. The method of claim 38 in which the second condition is that a counter is above a limit.

41. The method of claim 38 in which redirecting processing to the error-handling graph component also includes enabling a set of graph components, the set having been determined prior to the error.

42. A system for processing graph-based computations, the system including:

within a graph including vertices representing graph components that process work elements according to links joining the vertices,

means for providing at least one error-handling graph component configured to provide error information to a process external to the graph, and

means for processing data, including, in response to a graph component encountering an error while processing, redirecting processing to the error-handling graph component including directing at least some of the work elements to the error-handling component according to at least one link to a vertex representing the error-handling component.

43. A computer-readable medium storing a computer program for processing graph-based computations, the computer program including instructions for causing a computer system to:

within a graph including vertices representing graph components that process work elements according to links joining the vertices,

provide at least one error-handling graph component configured to provide error information to a process external to the graph, and

process data, including, in response to a graph component encountering an error while processing, redirecting processing to the error-handling graph component including directing at least some of the work elements to the error-handling component according to at least one link to a vertex representing the error-handling component.