Trigger point detection for online root cause analysis and system fault diagnosis
A computer-implemented method for detecting trigger points to identify root cause failure and fault events is provided. The method includes collecting, by a monitoring agent, entity metrics data and system key performance indicator (KPI) data, integrating the entity metrics data and the KPI data, constructing an initial system state space, detecting system state changes by calculating a distance between current batch data and an initial state, and dividing a system status into different states.
1. A computer-implemented method for detecting trigger points to identify root cause failure and fault events in a microservice system, the method comprising:
collecting, by a monitoring agent, entity metrics data and system key performance indicator (KPI) data from the microservice system;
integrating the entity metrics data and the KPI data;
constructing an initial system state space;
detecting system state changes by calculating a distance between current batch data and an initial state;
dividing a status of the microservice system into different states;
learning a causal graph over the system state changes to detect pods and nodes most likely to cause the root cause failure and fault events using a disentangle graph learning-based incremental discovery framework; and
mitigating damage to the microservice system from the failure and fault events using the detected pods and nodes.
2. The computer-implemented method of claim 1 , wherein the initial system state space is constructed by using a time-lagged trajectory matrix.
3. The computer-implemented method of claim 2 , wherein the time-lagged trajectory matrix leverages non-linear kernel-based matrix decomposition to learn a robust subspace.
4. The computer-implemented method of claim 2 , wherein, upon detection of a trigger point, the time-lagged trajectory matrix is updated.
5. The computer-implemented method of claim 1 , wherein a cumulative sum chart (CUSUM) is employed to accumulate a distribution error until a threshold is achieved.
6. The computer-implemented method of claim 1 , wherein the system state changes are represented as:
D ( t )=|| Û T @X ( t−L+ 1: t )|| F 2 −c
where @ refers to a matrix multiplication, X (t−L+1:t) is a new batch of data, c is an adjustable parameter, and Û T is a robust subspace.
7. A computer program product for detecting trigger points to identify root cause failure and fault events in a microservice system, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:
collecting, by a monitoring agent, entity metrics data and system key performance indicator (KPI) data from the microservice system;
integrating the entity metrics data and the KPI data;
constructing an initial system state space;
detecting system state changes by calculating a distance between current batch data and an initial state;
dividing a status of the microservice system into different states;
learning a causal graph over the system state changes to detect pods and nodes most likely to cause the root cause failure and fault events using a disentangle graph learning-based incremental discovery framework; and
mitigating damage to the microservice system from the failure and fault events using the detected pods and nodes.
8. The computer program product of claim 7 , wherein the initial system state space is constructed by using a time-lagged trajectory matrix.
9. The computer program product of claim 8 , wherein the time-lagged trajectory matrix leverages non-linear kernel-based matrix decomposition to learn a robust subspace.
10. The computer program product of claim 8 , wherein, upon detection of a trigger point, the time-lagged trajectory matrix is updated.
11. The computer program product of claim 7 , wherein a cumulative sum chart (CUSUM) is employed to accumulate a distribution error until a threshold is achieved.
12. The computer program product of claim 7 , wherein the system state changes are represented as:
D ( t )=|| Û T @X ( t−L+ 1: t )|| F 2 −c
where @ refers to a matrix multiplication, X (t−L+1:t) is a new batch of data, c is an adjustable parameter, and Û T is a robust subspace.
13. A computer processing system for detecting trigger points to identify root cause failure and fault events in a microservice system, comprising:
a memory device for storing program code; and
a processor device, operatively coupled to the memory device, for running the program code to:
collect, by a monitoring agent, entity metrics data and system key performance indicator (KPI) data from the microservice system;
integrate the entity metrics data and the KPI data;
construct an initial system state space;
detect system state changes by calculating a distance between current batch data and an initial state;
divide a status of the microservice system into different states;
learn a causal graph over the system state changes to detect pods and nodes most likely to cause the root cause failure and fault events using a disentangle graph learning-based incremental discovery framework; and
mitigate damage to the microservice system from the failure and fault events using the detected pods and nodes.
14. The computer processing system of claim 13 , wherein the initial system state space is constructed by using a time-lagged trajectory matrix.
15. The computer processing system of claim 14 , wherein the time-lagged trajectory matrix leverages non-linear kernel-based matrix decomposition to learn a robust subspace.
16. The computer processing system of claim 14 , wherein, upon detection of a trigger point, the time-lagged trajectory matrix is updated.
17. The computer processing system of claim 13 , wherein a cumulative sum chart (CUSUM) is employed to accumulate a distribution error until a threshold is achieved.