Vmware, Inc.

Ashot Nshan Harutyunyan

Arnak Poghosyan

Naira Movses Grigoryan

Mazda A Marvasti

Nicholas Kushmerick

Clement Pang

Tigran Bunarjyan

Yanislav Yankov

Nelli Aghajanyan

George Oganesyan

Vahe Khachikyan

Lilit Harutyunyan

Vardan Movsisyan

Dev A Nag

Sam Israelyan

Astghik Grigoryan

Davit Baghdasaryan

Marine Harutyunyan

Vahan Tadevosyan

Harutyun Beybutyan

Nina Karapetyan

Vardan Hovhannisyan

Khachatur Nazaryan

Ruzan Hovhannisyan

Meruzhan Kerobyan

Avetik Hovhannisyan

Vaghinak Saghatelyan

Amak Poghosyan

Artur Grigoryan

Karen Aghajanyan

Tigran Avagimyants

Hovhannes Antonyan

Nshan Sharoyan

Karen Avagyan

The current document is directed to methods and systems that automatically generate training data for machine-learning-based components used by a metric-data processing-and-analysis component of a distributed computer system, a subsystem within a distributed computer system, or a standalone metric-data processing-and-analysis system. The training data sets are labeled using categorical KPI values. The machine-learning-based components are applied to metric data both for predicting anomalous operational behaviors and problems within the distributed computer system and for determination of potential causes of anomalous operational behaviors and problems within the distributed computer system. Training of machine-learning-based components is carried out concurrently and asynchronously with respect to other metric-data collection, aggregation, processing, storage, and analysis tasks.

Methods and systems that automatically predict distributed-computer-system performance degradation using automatically trained machine-learning components

Automated computer-implemented methods and systems for resolving performance problems with objects executing in a data center are described. The automated methods use machine learning to obtain rules defining relationships between probabilities of event types of in log messages and performance problems identified by a key performance indictor ('KPI') of the object. When a KPI violates a corresponding threshold, the rules are used to evaluate run time log messages that describe the probable root cause of the performance problem. An alert identifying the KPI threshold violation, and the log messages are displayed in a graphical user interface of an electronic display device.

Methods and systems for using machine learning to resolve performance problems with objects of a data center

Automated computer-implemented methods and systems for discovering clusters of alerts triggered by abnormal events occurring with objects in a data center are described. In one aspect, alerts with start times in a sliding run-time window are retrieved from an alerts database. Each alert corresponds to a run-time event occurring with an object of the data center. Clusters of alerts in the sliding run-time window are detected based on the start times of the alerts and topological proximity of the objects. High priority alerts in the clusters of alerts are determined based on alert types. The events associated with discovered clusters of alerts and high priority alerts are displayed in a graphical user interface ('GUI'). Time evolution clustering of alerts and coverage evolution of alerts are over time based on the start times of the alerts and topological proximity of objects exhibiting abnormal behavior in the data center.

Methods and systems for discovering incidents through clustering of alert occurring in a data center

Automated, computer-implemented methods and systems for resolving performance problems with objects executing in a data center are described. The automated methods use machine learning to train a model that comprises rules defining relationships between probabilities of event types of in log messages and values of a key performance indictor ('KPI') of the object over a historical time period. When a KPI violates a corresponding threshold, the rules are used to evaluate run time log messages that describe the probable root cause of the performance problem. An alert identifying the KPI threshold violation, and the log messages are displayed in a graphical user interface of an electronic display device.

Methods and systems for resolving performance problems with objects of a data center

Computer-implemented methods and systems described herein perform intelligent sampling of application traces generated by an application. Computer-implemented methods and systems determine different sampling rates based on frequency of occurrence of trace types and/or frequency of occurrence of durations of the traces. Each sampling rate corresponds to a different trace type and/or different duration. The sampling rates for low frequency trace types and durations are larger than the sampling rates for high frequency trace types and durations. The relatively larger sampling rates for low frequency trace types and low frequency durations ensures that low frequency trace types and low frequency durations are sampled in sufficient numbers and are not passed over during sampling of the application traces. The set of sampled traces are stored in a data storage device.

Methods and systems for intelligent sampling of application traces

Automated methods and systems for identifying and resolving performance problems of objects of a data center are described. The automated methods and systems construct a model for identifying objects of the datacenter that are experiencing performance problems based on baseline distributions of events of the objects in a historical time period and event distributions of events of the objects in a time window located outside the historical time period. A root causes and recommendations database is constructed for resolving performance problems based on remedial measures previously performed for resolving performance problems. The model is used to monitor the objects of data center for runtime performance problems. When a performance problem with an object is detected, the root causes and recommendations database is used to identify a root cause of the performance problem and generate a recommendation for resolving the performance problem in near real time.

Methods and systems for identifying and resolving root causes of performance problems in data center object

The current document is directed to methods and systems that employ call traces collected by one or more call-trace services to generate call-trace-classification rules to facilitate root-cause analysis of distributed-application operational problems and failures. In a described implementation, a set of automatically labeled call traces is partitioned by the generated call-trace-classification rules. Call-trace-classification-rule generation is constrained to produce relatively simple rules with greater-than-threshold confidences and coverages. The call-trace-classification rules may point to particular services and service failures, which provides useful information to distributed-application and distributed-computer-system managers and administrators attempting to diagnose operational problems and failures that arise during execution of distributed applications within distributed computer systems. Call-trace-classification rules that are useful in multiple diagnoses are maintained as diagnosis tools for future diagnoses.

Automated methods and systems that facilitate root cause analysis of distributed-application operational problems and failures

The current document is directed to methods and systems that employ call traces collected by one or more call-trace services to generate call-trace-classification rules to facilitate root-cause analysis of distributed-application operational problems and failures. In a described implementation, a set of automatically labeled call traces is partitioned by the generated call-trace-classification rules. Call-trace-classification-rule generation is constrained to produce relatively simple rules with greater-than-threshold confidences and coverages. The call-trace-classification rules may point to particular services and service failures, which provides useful information to distributed-application and distributed-computer-system managers and administrators attempting to diagnose operational problems and failures that arise during execution of distributed applications within distributed computer systems. A first dataset is collected during normal distributed-application operation and a second dataset is collected during problem-associated or failure-associated operation of the distributed application. The first and second datasets are used to generate noise-subtracted call-trace-classification rules and/or diagnostic suggestions.

Automated methods and systems that facilitate root-cause analysis of distributed-application operational problems and failures by generating noise-subtracted call-trace-classification rules

Automated methods and systems for identifying problems associated with objects of a data center are described. Automated methods and systems are performed by an operations management server. For each object, the server determines a baseline distribution from historical events that are associated with a normal operational state of an object. The server determines a runtime distribution of runtime events that are associated with the object and detected in a runtime window of the object. The management server monitors runtime performance of the object while the object is running in the datacenter. When a performance problem is detected, the management server determines a root cause of a performance problem based on the baseline distribution and the runtime distribution and displays an alert in a graphical user interface of a display.

Automated methods and systems for identifying problems in data center objects

Automated processes and systems troubleshoot and optimize performance of applications running in distributed computing systems. An automated computer-implemented processes train an inference model for an application based on metrics associated with the application and a key performance indicator ('KPI') of the application. When a run-time performance problem is detected in run-time KPI values of KPI, the trained inference model is applied to run-time metrics and run-time KPI values to identify relevant run-time metrics that can be used to identify the root cause of the performance problem. The root cause of the performance problem can be used to generate a recommendation for correcting the performance problem. An alert identifying the root cause of the performance problem and the recommendation for correcting the performance problem are displayed on an interface of a display, thereby enabling correction of the performance problem and optimization of the application.

Automated methods and systems for troubleshooting and optimizing performance of applications running in a distributed computing system

Automated processes and systems for detecting abnormally behaving objects of a distributed computing system are described. Processes and systems obtain metrics that are generated in a historical time window and are associated with an object of the distributed computing system. Processes and system use the metrics to compute a time-dependent system indicator over the historical time window. Each value of the system indicator corresponds to a point in time of the historical time window when the object was in a normal or an abnormal state. Processes and systems use the normal and abnormal states of the system indicator in the historical time window to train a state classifier that is used to detect run-time abnormal behavior of the object. When the state classifier identifies abnormal behavior of the object, an alert is generated, indicating the abnormal behavior of the object.

Processes and systems that detect abnormal behavior of objects of a distributed computing system

The current document is directed to methods and systems that employ distributed-computer-system metrics collected by one or more distributed-computer-system metrics-collection services, call traces collected by one or more call-trace services, and attribute values for distributed-computer-system components to identify attribute dimensions related to anomalous behavior of distributed-computer-system components. In a described implementation, nodes correspond to particular types of system components and node instances are individual components of the component type corresponding to a node. Node instances are associated with attribute values and node are associated with attribute-value spaces defined by attribute dimensions. A set of call traces is partitioned, by clustering. Using attribute values and call traces, attribute dimensions that are likely related to particular anomalous behaviors of distributed-computer-system components are determined by decision-tree-related analyses for each partition and are reported to one or more computational entities to facilitate resolution of the anomalous behaviors.

Methods and systems that identify dimensions related to anomalies in system components of distributed computer systems using clustered traces, metrics, and component-associated attribute values

Automated computational methods and systems to classify and troubleshoot problems in information technology ('IT') systems or services provided by a distributed computing system are described. Each IT system of the distribution computing system or IT service provided by the distributed computing system has an associated key performance indicator ('KPI') used to monitor performance of the IT system or service. When real-time KPI data violates a KPI threshold, a real-time event-type distribution is computed from event messages generated by event sources associated with the IT system or service following the threshold violation. The real-time event-type distribution is compared with historical event-type distributions recorded for the KPI data in order to identify the problem and execute remedial action to resolve the problem.

Automated methods and systems to classify and troubleshoot problems in information technology systems and services

Methods and systems described herein are directed to troubleshooting anomalous behavior in a data center. Anomalous behavior in an object of a data center, such as a computational resource, an application, or a virtual machine ('VM'), may be related to the behavior of other objects at different hierarchies of the data center. Methods and systems provide a graphical user interface that enables a user to select a selected metric associated with an object of the data center experiencing a performance problem. Unexpected metrics of an object topology of the data center that correspond to the performance problem are identified. A recommendation for executing remedial measures to correct the performance problem is generated based on the unexpected metrics.

Methods and systems for troubleshooting anomalous behavior in a data center

Automated methods and systems to determine a baseline event-type distribution of an event source and use the baseline event type distribution to detect changes in the behavior of the event source are described. In one implementation, blocks of event messages generated by the event source are collected and an event-type distribution is computed for each of block of event messages. Candidate baseline event-type distributions are determined from the event-type distributions. The candidate baseline event-type distribution has the largest entropy of the event-type distributions. A normal discrepancy radius of the event-type distributions is computed from the baseline event-type distribution and the event-type distributions. A block of run-time event messages generated by the event source is collected. A run-time event-type distribution is computed from the block of run-time event messages. When the run-time event-type distribution is outside the normal discrepancy radius, an alert is generated indicating abnormal behavior of the event source.

Growing community of inventors

Yerevan, Armenia

Ashot Nshan Harutyunyan