TELEMETRY DATA CONTEXTUALIZED ACROSS DATASETS

• US 20170364561A1
• Filed: 06/20/2016
• Published: 12/21/2017
• Est. Priority Date: 06/20/2016
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Wu, Yi-Lei, Mihalcin, Adam Kenneth, Lopez Gilsinia Gilroy
• Granted Patent: US 10,540,358 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/2456 Join operations
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: H04L 67/125 involving control of end-device applications over a network

Abstract

Systems, methods, and software for telemetry event correlation is provided herein. An exemplary method includes obtaining an indication of a data selection defining at least one data entry of interest among datasets that comprise data entries determined by one or more associated telemetry elements, establishing a joined dataset based at least on contextually corresponding characteristics derived from the datasets, and processing the joined dataset with a plurality of adaptive functions to correlate data entries of the joined dataset to the data selection, with each of the plurality of adaptive functions configured to produce an associated correlation dataset comprising data entries potentially related to the data selection. The method also includes evaluating each associated correlation dataset to select an output dataset comprising contextual data entries related to the data selection.

Claims

1. A method of correlating data elements among telemetry datasets, the method comprising:

obtaining an indication of a data selection defining at least one data entry of interest among datasets that comprise data entries determined by one or more associated telemetry elements;

establishing a joined dataset based at least on contextually corresponding characteristics derived from the datasets;

processing the joined dataset with a plurality of adaptive functions to correlate data entries of the joined dataset to the data selection, with each of the plurality of adaptive functions configured to produce an associated correlation dataset comprising data entries potentially related to the data selection; and

evaluating each associated correlation dataset to select an output dataset comprising contextual data entries related to the data selection.

2. The method of claim 1, wherein the contextually corresponding characteristics comprise a time correlation among the datasets, and further comprising:

establishing the joined dataset based at least on the time correlation among the datasets, the time correlation including a timeframe associated with the data selection.

3. The method of claim 2, further comprising:

establishing the time correlation among the datasets by at least determining time properties associated with the data entries of the datasets; and

forming the joined dataset by merging portions of the datasets having corresponding time properties.

4. The method of claim 1, wherein the data entries of the datasets are not related by data key associations, and wherein the contextually corresponding characteristics are derived from at least properties of the datasets.

5. The method of claim 1, further comprising:

receiving an indication of a first noted event with the data selection, the first noted event associated with at least a first dataset; and

for each of the plurality of adaptive functions, determining potential events among the joined dataset that are evaluated by the associated adaptive function to correspond to a context of the first noted event.

6. The method of claim 5, further comprising:

correlating the first noted event to a target issue associated with a monitored element among the one or more telemetry elements;

for each of the plurality of adaptive functions, identifying a pattern of operational events that indicate the target issue among the joined dataset.

7. The method of claim 1, further comprising:

for each of the plurality of adaptive functions, processing the joined dataset with one or more machine learning algorithms to determine correlations of data entries in the joined dataset to the at least one data entry of interest and produce the associated correlation dataset.

8. The method of claim 7, further comprising:

evaluating each associated correlation dataset to select the output dataset based at least on a ranking of performance indicators associated with the one or more machine learning algorithms having processed past datasets of a similar type as one or more of the datasets.

9. The method of claim 1, further comprising:

selecting ones of the plurality of adaptive functions to process the joined dataset based on at least one selection factor comprising properties of data indicated by the data selection, properties of the data entries determined by the one or more associated telemetry elements, and indications of performance of the plurality of adaptive functions with regards to further datasets that share characteristics with the datasets.

10. An apparatus comprising:

one or more computer readable storage media;

program instructions stored on the one or more computer readable storage media that, when executed by a processing system, direct the processing system to at least;

obtain an indication of a data selection defining at least one data entry of interest among datasets that comprise data entries determined by one or more associated telemetry elements;

establish a joined dataset based at least on contextually corresponding characteristics derived from the datasets;

process the joined dataset with a plurality of adaptive functions to correlate data entries of the joined dataset to the data selection, with each of the plurality of adaptive functions configured to produce an associated correlation dataset comprising data entries potentially related to the data selection; and

evaluate each associated correlation dataset to select an output dataset comprising contextual data entries related to the data selection.

11. The apparatus of claim 10, wherein the contextually corresponding characteristics comprise a time correlation among the datasets, and comprising further program instructions, when executed by the processing system, direct the processing system to at least:

establish the joined dataset based at least on the time correlation among the datasets, the time correlation including a timeframe associated with the data selection.

12. The apparatus of claim 11, comprising further program instructions, when executed by the processing system, direct the processing system to at least:

establishing the time correlation among the datasets by at least determining time properties associated with the data entries of the datasets; and

forming the joined dataset by merging portions of the datasets having corresponding time properties.

13. The apparatus of claim 10, wherein the data entries of the datasets are not related by data key associations, and wherein the contextually corresponding characteristics are derived from at least properties of the datasets.

14. The apparatus of claim 10, comprising further program instructions, when executed by the processing system, direct the processing system to at least:

receive an indication of a first noted event with the data selection, the first noted event associated with at least a first dataset; and

for each of the plurality of adaptive functions, determine potential events among the joined dataset that are evaluated by the associated adaptive function to correspond to a context of the first noted event.

15. The apparatus of claim 14, comprising further program instructions, when executed by the processing system, direct the processing system to at least:

correlate the first noted event to a target issue associated with a monitored element among the one or more telemetry elements;

for each of the plurality of adaptive functions, identify a pattern of operational events that indicate the target issue among the joined dataset.

16. The apparatus of claim 10, comprising further program instructions, when executed by the processing system, direct the processing system to at least:

for each of the plurality of adaptive functions, process the joined dataset with one or more machine learning algorithms to determine correlations of data entries in the joined dataset to the at least one data entry of interest and produce the associated correlation dataset.

17. The apparatus of claim 16, comprising further program instructions, when executed by the processing system, direct the processing system to at least:

evaluate each associated correlation dataset to select the output dataset based at least on a ranking of performance indicators associated with the one or more machine learning algorithms having processed past datasets of a similar type as one or more of the datasets.

18. The apparatus of claim 10, comprising further program instructions, when executed by the processing system, direct the processing system to at least:

select ones of the plurality of adaptive functions to process the joined dataset based on at least one selection factor comprising properties of data indicated by the data selection, properties of the data entries determined by the one or more associated telemetry elements, and indications of performance of the plurality of adaptive functions with regards to further datasets that share characteristics with the datasets.

19. A method of operating a telemetry data processing system, the method comprising:

receiving an indication of event of interest that is included among datasets comprising operational events determined by one or more associated telemetry elements;

establishing a joined dataset from at least portions of the datasets that include a timeframe of the event of interest;

processing the joined dataset with a plurality of machine learning algorithms to correlate events of the joined dataset to the event of interest, with each of the plurality of machine learning algorithms configured to produce an associated dataset comprising data entries potentially related to the event of interest; and

evaluating each associated dataset to select an output dataset comprising contextual data entries accumulated from the datasets that are related to the event of interest.

20. The method of claim 19, further comprising:

evaluating each associated dataset to select the output dataset based at least on a ranking of performance indicators associated with the plurality of machine learning algorithms having processed past datasets of a similar type as one or more of the datasets.