Use of metrics selected based on lag correlation to provide leading indicators of service performance degradation
View Patent ↗The present description refers to a computer implemented method, computer program product, and computer system for identifying a service metric associated with a service, identifying one or more abnormalities of one or more infrastructure metrics that occur within a time window around an abnormality of the service metric, determining a set of candidate infrastructure metrics for the service metric based on how many times an abnormality of an infrastructure metric occurred within a time window around an abnormality of the service metric, determining a degree of lag correlation for each candidate infrastructure metric with respect to the service metric, selecting one or more candidate infrastructure metrics having a degree of lag correlation that exceeds a threshold to be a leading indicator infrastructure metric for the service metric, and providing a performance degradation warning for the service when an abnormality of one of the leading indicator infrastructure metrics is detected.
1. A computer program product, the computer program product being tangibly embodied on a non-transitory computer-readable storage medium and including executable code that, when executed, is configured to cause at least one data processing apparatus to:
identify a service metric associated with a service;
identify one or more abnormalities of one or more infrastructure metrics that occur within a time window around an abnormality of the service metric;
determine a set of candidate infrastructure metrics for the service metric based on how many times an abnormality of an infrastructure metric occurred within a time window around an abnormality of the service metric;
determine a degree of lag correlation for each candidate infrastructure metric with respect to the service metric, by:
collecting a set of data points for each candidate infrastructure metric,
collecting a set of data points for the service metric, and
determining a degree of correlation by calculating a Pearson's correlation coefficient for the candidate infrastructure metric with respect to the service metric based on the collected sets of data points;
select one or more candidate infrastructure metrics having a degree of lag correlation that exceeds a threshold to be a leading indicator infrastructure metric for the service metric; and
automatically provide a performance degradation warning for the service when an abnormality of one of the leading indicator infrastructure metrics is detected.
2. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to identify one or more abnormalities of one or more infrastructure metrics that occur within a time window around an abnormality of the service metric comprises code that is configurable to cause at least one data processing apparatus to:
identify an abnormality of the service metric; and
identify one or more abnormalities of one or more infrastructure metrics within a time window before and after the service metric abnormality.
3. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to identify one or more abnormalities of one or more infrastructure metrics that occur within a time window around an abnormality of the service metric comprises code that is configurable to cause at least one data processing apparatus to:
identify an abnormality of the service metric; and
identify one or more abnormalities of one or more infrastructure metrics within a time window preceding the service metric abnormality.
4. The computer program product of claim 3 wherein the code configured to cause at least one data processing apparatus to identify an abnormality of the service metric comprises the code configured to cause at least one data processing apparatus to:
collect data points for the service metric;
determine a baseline for the service metric based on the collected data points, the baseline including a high value and a low value;
identify an abnormality of the service metric by identifying any data points of the service metric that are outside of the baseline.
5. The computer program product of claim 3 wherein the code configured to cause at least one data processing apparatus to identify one or more abnormalities of one or more infrastructure metrics within a time window preceding the service metric abnormality comprises code configured to:
collect data points for one or more infrastructure metrics;
determine a baseline for each of the infrastructure metrics based on the collected data points, each baseline including a high value and a low value;
identify an abnormality of each of the infrastructure metrics by identifying any data points of the infrastructure metrics that are outside of the baseline for the infrastructure metric.
6. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to determine a set of candidate infrastructure metrics for the service metric comprises the code configured to cause at least one data processing apparatus to:
determine, for each of a plurality of infrastructure metrics, a number of times that an abnormality of an infrastructure metric occurred during a time window before an abnormality of the service metric; and
determine a set of candidate infrastructure metrics to include those infrastructure metrics that had the highest number of abnormalities within a time window before one or more service metric abnormalities as measured over a time period.
7. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to determine a degree of lag correlation by calculating a Pearson' correlation coefficient comprises code configured to:
calculate at least a first Pearson's correlation coefficient based on the sets of data points wherein one of the sets of data points is lagged by a first lag amount and a second Pearson's correlation coefficient based on the sets of data points wherein one of the sets of data points is lagged by a second lag amount; and
select either the first Pearson's correlation coefficient or the second Pearson's correlation coefficient, whichever is higher, as the degree of correlation for the candidate infrastructure metric.
8. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to determine a degree of lag correlation for a candidate infrastructure metric with respect to the service metric comprises code configured to:
collect a set of data points for a candidate infrastructure metric;
collect a set of data points for the service metric;
determine a degree of correlation for the candidate infrastructure metric by calculating a Pearson's correlation coefficient for the candidate infrastructure metric with respect to the service metric based on the collected data points for the service metric and a lagged set of data points for the candidate infrastructure metric.
9. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to determine a degree of lag correlation for a candidate infrastructure metric with respect to the service metric comprises code configured to:
collect a set of data points for a candidate infrastructure metric;
collect a set of data points for the service metric;
determine a degree of correlation for the candidate infrastructure metric by calculating a Pearson's correlation coefficient for the candidate infrastructure metric with respect to the service metric based on the collected data points for the candidate infrastructure metric and a lagged set of data points for the service metric.
10. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to provide a performance degradation warning comprises code configured to cause a data processing apparatus to:
detect an abnormality of one of the candidate infrastructure metrics;
provide a performance degradation warning for the service based on the detection of the abnormality of one of the candidate infrastructure metrics.
11. The computer program product of claim 1 wherein the code configured to cause at least one data processing apparatus to provide a performance degradation warning comprises code configured to cause a data processing apparatus to provide a performance degradation warning for the service based on the detection of the abnormality of one of the candidate infrastructure metrics, the performance degradation warning identifying the infrastructure metric, an abnormality that occurred for a triggering infrastructure metric that triggered providing the performance degradation warning for the service, and a lag amount that identifies the lag amount associated with the highest correlation coefficient for the triggering candidate infrastructure metric.
12. A computer implemented method comprising:
identifying a service metric associated with a service;
identifying one or more abnormalities of one or more infrastructure metrics that occur within a time window around an abnormality of the service metric;
determining a set of candidate infrastructure metrics for the service metric based on how many times an abnormality of an infrastructure metric occurred within a time window around an abnormality of the service metric;
determining a degree of lag correlation for each candidate infrastructure metric with respect to the service metric by:
collecting a set of data points for each candidate infrastructure metric;
collecting a set of data points for the service metric; and
determining a degree of correlation by calculating a Pearson's correlation coefficient for the candidate infrastructure metric with respect to the service metric based on the collected sets of data points, wherein one of the sets of data are lagged by a lag amount;
selecting one or more candidate infrastructure metrics having a degree of lag correlation that exceeds a threshold to be a leading indicator infrastructure metric for the service metric; and
automatically providing a performance degradation warning for the service when an abnormality of one of the leading indicator infrastructure metrics is detected.
13. The computer implemented method of claim 12 wherein the identifying one or more abnormalities of one or more infrastructure metrics that occur within a time window around an abnormality of the service metric comprises:
identifying an abnormality of the service metric; and
identifying one or more abnormalities of one or more infrastructure metrics within a time window before the service metric abnormality.
14. The computer implemented method of claim 13 wherein the identifying an abnormality of the service metric comprises:
collecting data points for the service metric;
determining a baseline for the service metric based on the collected data points;
identifying an abnormality of the service metric by identifying any data points of the service metric that are outside of the baseline.
15. The computer implemented method of claim 13 wherein the identifying one or more abnormalities of one or more infrastructure metrics within a time window preceding the service metric abnormality comprises:
collecting data points for one or more infrastructure metrics;
determining a baseline for each of the infrastructure metrics based on the collected data points; and
identifying an abnormality of each of the infrastructure metrics by identifying any data points of the infrastructure metrics that are outside of the baseline for the infrastructure metric.
16. The computer implemented method of claim 12 wherein the providing a performance degradation warning comprises:
detecting an abnormality of one of the candidate infrastructure metrics;
providing a performance degradation warning for the service based on the detection of the abnormality of one of the candidate infrastructure metrics.
17. An apparatus comprising:
a processor executing:
service metric identification logic configured to identify a service metric associated with a service;
abnormality identification logic configured to identify one or more abnormalities of one or more infrastructure metrics that occur within a time window around an abnormality of the service metric;
determining logic configured to determine a set of candidate infrastructure metrics for the service metric based on how many times an abnormality of an infrastructure metric occurred within a time window around an abnormality of the service metric;
the determining logic configured to determine a degree of lag correlation for each candidate infrastructure metric with respect to the service metric by:
collecting a set of data points for each candidate infrastructure metric;
collecting a set of data points for the service metric; and
determining a degree of correlation by calculating a Pearson's correlation coefficient for the candidate infrastructure metric with respect to the service metric based on the collected sets of data points, wherein one of the sets of data points is lagged by a lag amount;
selecting logic configured to select one or more candidate infrastructure metrics having a degree of lag correlation that exceeds a threshold to be a leading indicator infrastructure metric for the service metric; and
warning providing logic configured to automatically provide a performance degradation warning for the service when an abnormality of one of the leading indicator infrastructure metrics is detected.