Correlation enhanced causation identification for data processing system management
Methods and systems for managing customer-encountered issues are disclosed. To manage the customer-encountered issues, predictions regarding the causes of customer-encountered issues may be obtained and used to guide remediation processes. The predictions may be obtained using inference models that utilize correlations to filter potential causes of the issues. By filtering the causes, the resulting predictions may be more likely to be accurate thereby improving the rate at which customer-encountered issues may be resolved.
1 . A method for managing customer-encountered issues, the method being performed by a hardware processor of a response management system (RMS) and comprising:
identifying an occurrence of an issue for a data processing system managed by the RMS;
based on the identifying of the occurrence:
obtaining a first ordered set of events associated with the issue;
using a knowledge base and the first ordered set of events to obtain a second ordered set of events that have been previously encountered in remediations of previously encountered issues;
obtaining pairing statistics associated with the issue;
using a knowledge graph and the pairing statistics to obtain a first structure that has been previously encountered in the remediations of the previously encountered issues;
ingesting the second ordered set of events and the first structure into an inference model hosted by the RMS to generate, as an output of the inference model, a second structure for the issue; and
using the second structure to remediate the issue for the data processing system to obtain a remediated one of the data processing system.
2 . The method of claim 1 , wherein the inference model comprises a transformer model trained to determine and output, at least, a cause of the issue.
3 . The method of claim 2 , wherein the second structure comprises the cause of the issue and an effect of the cause.
4 . The method of claim 1 , wherein the first ordered set of events comprise events that occurred with respect to the data processing system leading up to the issue.
5 . The method of claim 4 , wherein the first ordered set of events are obtained using, at least in part, logs of operation of the data processing system.
6 . The method of claim 5 , wherein the pairing statistics comprise at least one precondition event for the issue to occur.
7 . The method of claim 1 , wherein the knowledge base comprises previously encountered ordered sets of events for corresponding previously encountered issues.
8 . The method of claim 7 , wherein the knowledge graph comprises known cause and effect relationships for the previously encountered issues.
9 . The method of claim 1 , wherein the second structure comprises a three-dimensional space.
10 . The method of claim 9 , wherein the three-dimensional space comprises a first dimension that maps to temporal relationships, a second dimension that maps to statistical correlation relationships, and a third dimension that maps to cause and effect relationships.
11 . A non-transitory machine-readable medium having instructions stored therein, which when executed by a hardware processor of a response management system (RMS), cause the hardware processor to perform operations for managing customer-encountered issues, the operations comprising:
identifying an occurrence of an issue for a data processing system managed by the RMS;
based on the identifying of the occurrence:
obtaining a first ordered set of events associated with the issue;
using a knowledge base and the first ordered set of events to obtain a second ordered set of events that have been previously encountered in remediations of previously encountered issues;
obtaining pairing statistics associated with the issue;
using a knowledge graph and the pairing statistics to obtain a first structure that has been previously encountered in the remediations of the previously encountered issues;
ingesting the second ordered set of events and the first structure into an inference model hosted by the RMS to generate, as an output of the inference model, a second structure for the issue; and
using the second structure to remediate the issue for the data processing system to obtain a remediated one of the data processing system.
12 . The non-transitory machine-readable medium of claim 11 , wherein the inference model comprises a transformer model trained to determine and output, at least, a cause of the issue.
13 . The non-transitory machine-readable medium of claim 12 , wherein the second structure comprises the cause of the issue and an effect of the cause.
14 . The non-transitory machine-readable medium of claim 11 , wherein the first ordered set of events comprise events that occurred with respect to the data processing system leading up to the issue.
15 . The non-transitory machine-readable medium of claim 14 , wherein the first ordered set of events are obtained using, at least in part, logs of operation of the data processing system.
16 . A response management system (RMS), comprising:
a hardware processor; and
a memory coupled to the hardware processor to store instructions, which when executed by the hardware processor, cause the hardware processor to perform operations for managing customer-encountered issues, the operations comprising:
identifying an occurrence of an issue for a data processing system managed by the RMS;
based on the identifying of the occurrence:
obtaining a first ordered set of events associated with the issue;
using a knowledge base and the first ordered set of events to obtain a second ordered set of events that have been previously encountered in remediations of previously encountered issues;
obtaining pairing statistics associated with the issue;
using a knowledge graph and the pairing statistics to obtain a first structure that has been previously encountered in the remediations of the previously encountered issues;
ingesting the second ordered set of events and the first structure into an inference model hosted by the RMS to generate, as an output of the inference model, a second structure for the issue; and
using the second structure to remediate the issue for the data processing system to obtain a remediated one of the data processing system.
17 . The system of claim 16 , wherein the inference model comprises a transformer model trained to determine and output, at least, a cause of the issue.
18 . The system of claim 17 , wherein the second structure comprises the cause of the issue and an effect of the cause.
19 . The system of claim 16 , wherein the first ordered set of events comprise events that occurred with respect to the data processing system leading up to the issue.
20 . The system of claim 19 , wherein the first ordered set of events are obtained using, at least in part, logs of operation of the data processing system.