Interactive detection of system anomalies

1. A system comprising:

a processor;

a non-transitory computer readable medium on which is stored instructions that when executed by the processor, cause the processor to:

receive input data related to a series of events, each event among the series of events being associated with a respective event type in a reference group of event types;

generate an anomaly score based on a number of each event type in the series of events and a rarity of each event type with respect to other event types in the reference group of event types;

detect presence of a system anomaly based on the anomaly score;

compare the series of events of the system anomaly to previous series of events of previous system anomalies that were presented to users and for which interactions were received from the users, wherein the interactions indicate whether or not the previous system anomalies were of interest to the users;

determine that the system anomaly is of potential interest based on the comparison;

generate, to be displayed via an interactive graphical user interface, an output data stream based on the determination that the system anomaly is of potential interest;

receive, from the interactive graphical user interface, feedback data associated with the output data stream; and

store the feedback data as an interaction for comparison with subsequent system anomalies.

2. The system of claim 1 , wherein the system anomaly is associated with a time, and wherein the instructions when executed by the processor, further cause the processor to: determine, for a time interval, at least one of an anomaly intensity, the anomaly score, an anomaly fingerprint, and a fingerprint matching function.

3. The system of claim 2 , wherein the instructions when executed by the processor, further cause the processor to: detect a presence of the system anomaly based on the anomaly fingerprint, and the fingerprint matching function.

4. The system of claim 2 , wherein the instructions when executed by the processor, further cause the processor to:

determine, for the time interval, anomaly intensities, including the anomaly intensity, and the anomaly score, and wherein each anomaly intensity from among the anomaly intensities is to be transformed, with respect to a distribution of anomaly intensities of an anomaly type in reference time-slots, based on a distinctive residual rarity extremity score, into comparable, additive, and distinctive anomaly intensity scores that are combined to determine the anomaly score.

5. The system of claim 2 , wherein each event in the series of events is associated with an event type, a time, and zero or more measurement values, and wherein the instructions when executed by the processor, further cause the processor to:

determine, for each event type, an anomaly intensity amount for an anomaly type from events in the time interval, wherein for each anomaly type, the anomaly intensity amounts for different event types are combined to determine the anomaly intensity and the anomaly fingerprint, and wherein the anomaly fingerprint is based on a set of relative contributions of event types to the anomaly intensity.

6. The system of claim 5 , wherein the instructions when executed by the processor, further cause the processor to:

determine, for the time interval for an anomaly type, incomparable anomaly intensity amounts, wherein each incomparable anomaly intensity amount is transformed with respect to a distribution of associated incomparable anomaly intensity amounts in reference time intervals, based on a distinctive residual rarity extremity score, into comparable, additive, and distinctive anomaly intensity amounts.

7. The system of claim 5 , wherein the anomaly type is selected from among:

a Flood of Events, wherein the anomaly intensity amount is an event count,

a Variety of Events, wherein the anomaly intensity amount is an event occurrence indicator,

a Flood of Rare Events, wherein the anomaly intensity amount is a product of an event count extremity factor, and an event-type rarity factor, and

a Flood of Extreme Signals, wherein the anomaly intensity amount is a maximal signal value per time interval transformed based on a distinctive residual rarity extremity score.

8. The system of claim 5 , wherein the instructions when executed by the processor, further cause the processor to:

identify the anomaly fingerprint, and wherein a fingerprint matching score for the anomaly fingerprint is computed in a second time interval to determine presence or absence of similar system anomalies in the second time interval, the fingerprint matching score computed based on a correlation between the anomaly fingerprint and anomaly intensity amounts in the second time interval.

9. The system of claim 1 , wherein the instructions when executed by the processor, further cause the processor to:

generate an interactive analysis interface to be provided via the interactive graphical user interface, and modify the output data stream based on interactions with the interactive analysis interface.

10. The system of claim 9 , wherein the instructions when executed by the processor, further cause the processor to:

detect, based on the interactions with the interactive analysis interface, a system anomaly associated with a Partial Pattern anomaly type.

11. The system of claim 1 , wherein the instructions when executed by the processor, further cause the processor to:

detect event patterns based on at least one of the feedback data and the previous system anomalies.

12. The system of claim 1 , wherein the instructions when executed by the processor, further cause the processor to:

generate a word cloud to be displayed via the interactive graphical user interface, the word cloud highlighting words that appear in log messages associated with a selected system anomaly.

13. A method comprising:

receiving, by a processor, input data related to a series of events, each event in the series of events being associated with a respective event type in a reference group of event types;

generating, by the processor, for each event type in the reference group of event types, a respective component anomaly score based on a number of the event types in the series of events and a rarity of the event type with respect to other event types in the reference group of event types;

generating, by the processor, an anomaly score based on the respective component anomaly score generated for each event type in the reference group of events types;

generating, by the processor, an output data stream based on system anomalies detected based on the anomaly score and other anomaly scores;

generating, by the processor, to be displayed via an interactive graphical user interface, the output data stream including an attribute associated with the output data stream;

receiving, by the processor, from the interactive graphical user interface, feedback data indicative of selection of a system anomaly;

processing, by the processor, the feedback data to modify the output data stream;

providing, by the processor, via the interactive graphical user interface, an interactive analysis interface for operations analytics based on the selected system anomaly;

comparing, by the processor, a second system anomaly to the selected system anomaly;

determining, by the processor, that the second system anomaly is of potential interest based on the comparison; and

displaying, by the processor, via the interactive graphical user interface, the second system anomaly.

14. The method of claim 13 , wherein the attribute associated with the output data stream includes at least one of an anomaly intensity, the anomaly score, an anomaly fingerprint, a fingerprint matching function, event patterns, a word cloud, an anomaly type, a service message associated with the selected system anomaly, an anomaly intensity for events in a time interval, an event count extremity factor, and an event type rarity factor.

15. A non-transitory computer readable medium comprising executable instructions that when executed by a processor cause the processor to:

receive input data related to a series of events, each event in the series of events being associated with a respective event type in a reference group of event types;

generate an anomaly score based on a number of each event type in the series of events and a rarity of each event type with respect to other event types in the reference group of event types;

detect a system anomaly based on the anomaly score;

generate an output data stream based on the detected system anomaly;

provide the output data stream to be displayed via an interactive graphical user interface;

facilitate operations analytics of the series of events based on feedback data indicative of interactions with the output data stream, the interactions including selection of the system anomaly;

compare a second system anomaly to the selected system anomaly;

determine that the second system anomaly is of potential interest based on the comparison; and

display, via the interactive graphical user interface, the second system anomaly.

16. The non-transitory computer readable medium of 15 , wherein the input data comprises a plurality of unstructured messages related to the series of events, each of the unstructured messages having a corresponding format such that the input data has heterogeneous data formats, and wherein to detect the system anomaly, the instructions when executed by the processor further cause the processor to:

determine a set of structured message types and parameters based on the plurality of unstructured messages;

generate a regular expression for matching each of the plurality of unstructured messages based on the set of structured message types and parameters; and

identify an event type for each of the plurality of unstructured messages based on the regular expression.

17. The non-transitory computer readable medium of 15 , wherein the instructions when executed by the processor further cause the processor to:

identify a subset of the series of events in a time interval, wherein the selected system anomaly is detected based on the subset of the series of events in the time interval.

18. The non-transitory computer readable medium of 17 , wherein the instructions when executed by the processor further cause the processor to:

generate an anomaly fingerprint based on the identified subset of the series of events;

store the anomaly fingerprint;

generate a second anomaly fingerprint based on a second subset of the series of events, the second anomaly fingerprint being associated with the second system anomaly,

wherein the comparison of the second system anomaly to the selected system anomaly is based on a comparison of the anomaly fingerprint with the second anomaly fingerprint; and

determine that the second system anomaly is similar to the selected system anomaly based on the comparison, wherein the determination that the second system anomaly is of potential interest is based on the determination that the second system anomaly is similar to the selected system anomaly.

19. The non-transitory computer readable medium of 15 , wherein the instructions when executed by the processor further cause the processor to:

assign a second anomaly score to the second system anomaly, the second anomaly score indicative of a level of potential interest of the second system anomaly;

determine that the selected system anomaly was interacted with to obtain further information relating to the selected system anomaly; and

adjust the second anomaly score to indicate that the second system anomaly is of greater interest than before the determination that the selected system anomaly was interacted with to obtain further information relating to the selected system anomaly.

20. The non-transitory computer readable medium of 19 , wherein the instructions when executed by the processor further cause the processor to:

detect a third system anomaly based on the input data;

compare the third system anomaly to a set of previous system anomalies;

determine that the set of previous system anomalies were presented to users and that interactions with the set of previous system anomalies were not received from the users; and

determine that the third system anomaly is not of potential interest based on the comparison of the third system anomaly to the set of previous system anomalies and the determination that the set of previous system anomalies were presented to users and that interactions with the set of previous system anomalies were not received from the users.