System and method for troubleshooting abnormal behavior of an application
A method for troubleshooting abnormal behavior of an application hosted on a networked computer system. The method may be implemented by a root cause analyzer. The method includes tracking a single application performance metric across all the clients of an application hosted on a networked computer system and analyzing an aggregated application based on the single application metric. The method involves determining outlier client attributes associated with an abnormal transaction of the application and ranking the outlier client attributes based on comparisons of historical and current abnormal transactions. The method associates one or more of the ranked outlier client attributes with the root cause of the current abnormal transaction. Association rule learning is used to associate one or more of the ranked outlier client attributes with the root cause.
1. A method for root cause analysis of an application performance abnormality of an application in a networked computer system, the method comprising:
collecting client transactions data for transactions conducted by clients using the application on the networked computer system, the client transactions data including transaction records that include attributes and performance metrics of each transaction;
identifying at least one abnormal transaction from among the client transactions data based on a violation of a performance metric value of a performance metric of the performance metrics;
aggregating a subset of the client transactions data occurring within a time window of the at least one abnormal transaction;
filtering the subset of the client transactions data to identify a set of abnormal transactions that violate the performance metric value;
identifying a set of frequently appearing attributes that appear in the set of abnormal transactions;
identifying outlier attributes within the set of frequently appearing attributes by removing frequently appearing normal attributes occurring within a set of historical normal transactions data from the frequently appearing abnormal attributes; and
identifying a root cause for the abnormal event based on a frequency of occurrence of each of the outlier attributes within an aggregated set of abnormal transactions that includes the set of abnormal transactions and a set of historical abnormal transactions.
2. The method of claim 1 , wherein pinpointing a root cause event for the abnormal event includes ranking the outlier attributes based on the frequency of occurrence of each of the outlier attributes.
3. The method of claim 1 , wherein collecting client transactions data for transactions conducted by clients using the application on the networked computer system includes performing statistical functions including an average on the client transactions data over a period to establish baseline metrics data.
4. The method of claim 3 , further comprising: generating an abnormal metric event alert for a baseline violation by the metric.
5. The method of claim 1 , including collecting client transactions data for transactions conducted by clients using the application on the networked computer system and collecting data entities having a key-value structure that includes both contextual attributes and numeric metrics.
6. The method of claim 1 , wherein the performance metric value is determined based on aggregated performance metric values of the performance metric among the clients.
7. A system for root cause analysis of an application performance abnormality of an application in a networked computer system, the system comprising:
a computing device including at least one processor and at least one memory;
a monitor configured to collect client transactions data for transactions conducted by clients using the application on the networked computer system, the client transactions data including transaction records that include attributes and performance metrics of each transaction; and
an analyzer configured to:
identify at least one abnormal transaction from among the client transactions data based on a violation of a performance metric value of a performance metric of the performance metrics;
aggregate a subset of the client transactions data occurring within a time window of the at least one abnormal transaction;
filter the subset of the client transactions data to identify a set of abnormal transactions that violate the performance metric value;
identify a set of frequently appearing attributes that appear in the set of abnormal transactions;
identify outlier attributes within the set of frequently appearing attributes by removing frequently appearing normal attributes occurring within a set of historical normal transactions data from the frequently appearing abnormal attributes; and
identify a root cause for the abnormal event based on a frequency of occurrence of each of the outlier attributes within an aggregated set of abnormal transactions that includes the set of abnormal transactions and a set of historical abnormal transactions.
8. The system of claim 7 , wherein the analyzer is configured to rank the outlier attributes based on the frequency of occurrence of each of the outlier attributes.
9. The system of claim 7 , further comprising: an aggregator configured to perform statistical functions including an average on the client transactions data over a period of normal performance to establish baseline metrics data.
10. The system of claim 9 , wherein the aggregator is configured to generate an abnormal metric event alert for a baseline violation by the metric.
11. The system of claim 7 , wherein collecting client transactions data for transactions conducted by clients using the application on the networked computer system includes collecting data entities having a key-value structure including both contextual attributes and numeric metrics.
12. The system of claim 7 , wherein the performance metric value is determined based on aggregated performance metric values of the performance metric among the clients.
13. A computer program product including instructions recorded on a non-transitory computer-readable storage medium and configured to cause at least one processor to conduct root cause analysis of an application performance abnormality of an application in a networked computer system, the instructions when executed by the at least one processor causing the at least one processor to:
collect client transactions data for transactions conducted by clients using the application on the networked computer system, the client transactions data including transaction records that include attributes and performance metrics of each transaction;
identify at least one abnormal transaction from among the client transactions data based on a violation of a performance metric value of a performance metric of the performance metrics;
aggregate a subset of the client transactions data occurring within a time window of the at least one abnormal transaction;
filter the subset of the client transactions data to identify a set of abnormal transactions that violate the performance metric value;
identify a set of frequently appearing attributes that appear in the set of abnormal transactions;
identify outlier attributes within the set of frequently appearing attributes by removing frequently appearing normal attributes occurring within a set of historical normal transactions data from the frequently appearing abnormal attributes; and
identify a root cause event for the abnormal event based on a frequency of occurrence of each of the outlier attributes within an aggregated set of abnormal transactions that includes the set of abnormal transactions and a set of historical abnormal transactions.
14. The computer program product of claim 13 , wherein the instructions, when executed, cause the at least one processor to:
rank the outlier attributes based on the frequency of occurrence of each of the outlier attributes.
15. The computer program product of claim 13 , wherein the instructions, when executed, cause the at least one processor to:
average the client transactions data over a period of normal performance to establish baseline metrics data.
16. The computer program product of claim 14 , wherein the instructions, when executed, cause the at least one processor to:
generate an abnormal metric event alert for a baseline violation by the metric.
17. The computer program product of claim 13 , wherein collecting client transactions data for transactions conducted by clients using the application on the networked computer system involves collecting data entities having a key-value structure including both contextual attributes and numeric metrics.
18. The computer program product of claim 13 , wherein the performance metric value is determined based on aggregated performance metric values of the performance metric among the clients.