Granted Patent B2

US 12,608,261 · App. 18/127,711 · Granted Apr 21, 2026

Inventors: Larisa Shwartz (Greenwich, CT); Saurabh Jha (White Plains, NY); Jesus Maria Rios Aliaga (Philadelphia, PA); Eitan Daniel Farchi (Pardes Hanna-Karkur, IL); Frank Bagehorn (Dottikon, CH); Robert Filepp (Westport, CT)

Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATION

G06F11/0775G06F11/079

Abstract

Claims (41)

1 . A method, comprising:

determining, by a processor set, normal execution times of respective services in a call graph of an application;

determining, by the processor set, normal execution times of respective network communications between ones of the services;

determining, by the processor set, faulty execution times of respective ones of the services; and

generating, by the processor set, a propagation time for a particular type of fault injected at a particular fault injection location in the call graph based on the determined normal execution times of respective services, the determined normal execution times of respective network communications, and the determined faulty execution times of respective ones of the services,

wherein the faulty execution times of the respective ones of the services are determined by estimating a probability of a tail event representing a service level objective (SLO) violation using Extreme Value Theory (EVT).

2 . The method of claim 1 , where a fault is realized in different ways by optimization of combinatorial test design.

3 . The method of claim 1 , further comprising repeating the generating for each of plural different types of faults and each of plural different fault injection locations.

4 . The method of claim 3 , further comprising generating a list comprising multiple entries each comprising:

a respective one of the plural different types of faults;

a respective one of the plural different fault injection locations; and

the generated propagation time for a combination of the respective one of the plural different types of faults and the respective one of the plural different fault injection locations.

5 . The method of claim 4 , further comprising adjusting one or more entries of the list based on feedback.

6 . The method of claim 1 , wherein:

the application comprises hybrid cloud application that includes multiple microservices;

the services in the call graph correspond to respective ones of the multiple microservices and network communications; and

the network communications include directionality of execution between the respective ones of the multiple microservices and network.

7 . The method of claim 1 , wherein the normal execution times of the respective services are determined using tracing.

8 . The method of claim 7 , wherein the normal execution times of the respective services and network nodes comprise statistical distributions.

9 . The method of claim 1 , wherein the normal execution times of the respective network communications are determined using tracing and comprise statistical distributions.

10 . A method, comprising:

determining, by a processor set, normal execution times of respective services in a call graph of an application;

determining, by the processor set, normal execution times of respective network communications between ones of the services;

determining, by the processor set, faulty execution times of respective ones of the services; and

generating, by the processor set, a propagation time for a particular type of fault injected at a particular fault injection location in the call graph based on the determined normal execution times of respective services, the determined normal execution times of respective network communications, and the determined faulty execution times of respective ones of the services, wherein the generating the propagation time is further based on one or more parallelization numbers.

11 . A computer program product comprising one or more computer readable storage media having program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to:

determine normal execution times of respective services in a call graph of an application, wherein nodes of the call graph represent microservices in the application and paths between the nodes represent direction of request flow between the microservices;

determine normal execution times of respective network communications between ones of the services;

determine faulty execution times of respective ones of the services by running fault injection simulations; and

generate a propagation time for a particular type of fault injected at a particular fault injection location in the call graph based on the determined normal execution times of respective services, the determined normal execution times of respective network communications, and the determined faulty execution times of respective ones of the services,

wherein the program instructions are further executable to:

repeat the generating for each of plural different types of faults and each of plural different fault injection locations; and

generate a list comprising multiple entries each comprising:

a respective one of the plural different types of faults;

a respective one of the plural different fault injection locations; and

the generated propagation time for a combination of the respective one of the plural different types of faults and the respective one of the plural different fault injection locations; and

wherein the program instructions are further executable to:

generate training data based on the list; and

train a machine learning model using the training data, wherein the machine learning model is configured to predict a fault type and a fault location of the application based on an input comprising log data of the application.

12 . The computer program product of claim 11 , wherein:

the normal execution times of the respective services and the normal execution times of the respective network communications are determined using tracing.

Assignments (1)

Continuity (1)

References Cited (17)