US 12,326,864 B2
Method and system for operation objects discovery from operation data
Jia Qi Li, Beijing (CN); Fan Jing Meng, Beijing (CN); Pei Ni Liu, Beijing (CN); Junmei Qu, Beijing (CN); and Zi Xiao Zhu, Beijing (CN)
Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed by INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed on Mar. 15, 2020, as Appl. No. 16/819,143.
Prior Publication US 2021/0286819 A1, Sep. 16, 2021
Int. Cl. G06F 16/353 (2025.01); G06F 16/215 (2019.01); G06F 16/2455 (2019.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01); G06F 16/334 (2025.01)
CPC G06F 16/24558 (2019.01) [G06F 16/215 (2019.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01)] 17 Claims
OG exemplary drawing
 
1. A computer-implemented method of monitoring operation objects of a distributed workload multi-cloud computing platform, comprising:
receiving a workload;
distributing the received workload to the multi-cloud computing platform;
monitoring a real-time streaming operation data of the distributed workload multi-cloud computing platform by a pattern matching module of a computing device;
performing, by the pattern matching module of the computing device, pattern matching of the real-time streaming operation data with patterns in a database;
identifying each field in the operation data having matching patterns with the database as first potential objects;
performing data profiling on each of unmatched fields of the operation data to generate data profiles to profile the distribution of each field with various types of values, wherein the data profiling comprises analyzing the unmatched fields of the operation data and collecting statistics therefrom;
field classifying the data profiles to generate second potential objects based on the data profiles of the unmatched fields of the operation data, by a field classifier module operative to identify a particular class, a possible error pattern, and/or a warning pattern in the data profiles of the unmatched fields;
relieving a storage space of the multi-cloud computing platform by de-duplicating the first potential objects and the second potential objects by deleting duplicative data of the first potential objects and the second potential objects when generating operation objects in response to determining that at least two of the first potential objects and the second potential objects have a same value and different field names in different indices;
updating a knowledge base of the de-duplicated data of the first potential objects and the second potential objects; and
detecting at least one of segmented operations, errors, and/or faults in the distributed workload multi-cloud computing platform based on the generated operation objects.