Amazon Technologies, Inc.

Anton Vladilenovich Goldberg

Brandon William Porter

Daniel Wade Hitchcock

Aram Grigoryan

Darin Silva Nee

Sergei Rutenberg

Kyle Bradley Peterson

Michael James McInerny

Melissa Elaine Davis

Robert Hanna Mahfoud

Brian Thomas Smith

Trevor Alexander Popiel

Jon Arron McClintock

Harsha Ramalingam

Christopher Allen Suver

Ajit Ashok Varangaonkar

Aaron Ben Fernandes

Paul Derek DeMarco

Charles Alexander Carman

Donghui Zhuo

Srinivasa Susheel Kumar Mannepalli

Slavik Dimitrovich

Matthew Paul Wenger

James Daniel Longley

David Elfi

Michael David Decker

Renaud Bordelet

At a rule processing unit of an evolving, self-organized machine learning-based resource management service, a rule of a first rule set is applied to a value of a first collected metric, resulting in the initiation of a first corrective action. A set of metadata indicating the metric value and the corrective action is transmitted to a repository, and is used as part of an input data set for a machine learning model trained to generate rule modification recommendations. In response to determining that the corrective actions did not meet a success criterion, an escalation message is transmitted to another rule processing unit.

Artificial intelligence based self-organizing event-action management system for large-scale networks

Methods, systems, and computer-readable media for collecting route-based traffic metrics in a service-oriented system are disclosed. A first service in a service-oriented system receives an inbound request from a second service. From the inbound request, the first service extracts, a route identifier representing an upstream route including the second service. The first service stores the route identifier and a counter representing a number of inbound requests with the route identifier that have been received by the first service over a current window of time. The first service sends, to a traffic metric collection system, a metric message comprising the route identifier and the counter. Traffic metric data is determined by the traffic metric collection system using the metric message from the first service and one or more additional metric messages from one or more additional services. The traffic metric data indicates one or more flows of requests through the services.

Collecting route-based traffic metrics in a service-oriented system

Methods, systems, and computer-readable media for traffic distribution mapping in a service-oriented system are disclosed. A plurality of call paths are determined representing service interactions among a plurality of services. The call paths include a particular service and are determined using trace data generated by the services. Total call volumes are determined at individual ones of the services. Based at least in part on the call paths and the total call volumes, one or more estimated call ratios are determined between the particular service and one or more APIs of one or more additional services. Based at least in part on the call ratio(s) and the total call volumes, one or more call volumes are determined between the particular service and the one or more APIs of the one or more additional services.

Traffic distribution mapping in a service-oriented system

Methods, systems, and computer-readable media for implementing automated services capacity modeling using call tracing are disclosed. A plurality of demand drivers are determined based on trace data for service interactions between services in a service-oriented system. The demand drivers are determined to drive a generation of service calls to a particular service. A total call volume is determined to the particular service based on the external demand drivers. An optimized quantity of computing resources to provide the particular service is determined based on the total call volume.

Automated services capacity modeling

Methods, systems, and computer-readable media for real-time call tracing in a service-oriented system are disclosed. A request comprising a trace signature is received at a service in a service-oriented system. The trace signature comprises a representation of one or more upstream services in an upstream call path associated with the request. Based at least in part on the trace signature, the service determines one or more actions to perform with respect to the request.

Real-time call tracing in a service-oriented system

Methods, systems, and computer-readable media for implementing distributed trace storage in a service-oriented system are disclosed. A plurality of trace events are generated for a plurality of service interactions between individual ones of a plurality of services. The services are implemented by a plurality of hosts. Individual ones of the trace events comprise respective trace identifiers. The trace events are stored at individual ones of the hosts at which the respective trace events were generated. In response to a request from a trace analysis system, a portion of the trace events are retrieved and sent from individual ones of the hosts to the trace analysis system.

Distributed trace storage in a service-oriented system

Methods, systems, and computer-readable media for implementing pattern-based detection are disclosed. A plurality of services monitor a plurality of service interactions comprising data or metadata. The services compare the data or metadata to a set of patterns and identify one or more matched patterns among the set of patterns. The services send data indicative of the matched patterns to a central recording service. The central recording service aggregates the data indicative of the matched patterns and generates one or more data flow visualizations indicating one or more data flows between individual ones of the services for the matched patterns.

Pattern-based detection for services in distributed systems

A machine learning engine is configured to create a customized anomaly detector for use by a system resource, such as a virtual machine instance running specific operations. Anomalies are determined by comparing aspects of current data to 'normal' baseline data indicating a normal range of performance and operation of a system resource. Operation by a system resource that is outside of this normal range of performance and operation may be considered an anomaly. The machine learning engine may be used to create customized monitoring that detects anomalies in operation and/or performance of a system resource based on at least some custom parameters that are unique to the particular system that is to be monitored. The parameters may include operational parameters, which may be selected specifically for the system to be monitored. The parameters may also include some technical parameters which are used by many other systems to monitor hardware performance.

Anomaly detection using machine learning

Described are techniques for determining the effect of a change in how data is handled by a first device on other devices within a system. At a first device, sending, receiving, or processing of data may be prevented for a length of time. Metric data may be determined from other devices during the length of time and during one or more time periods before or after the length of time. The metric data may indicate the effect of the handling of data by a first device on the one or more other devices.

System for modeling distributed systems

Methods, systems, and computer-readable media for implementing interaction reconstruction in a service-oriented system are disclosed. Routes are stored for a first set of service interactions between individual services in a service-oriented system. The routes comprise data indicative of call paths for node identifiers in the first set of service interactions. A sparse trace is generated for a second set of service interactions between individual ones of the services. The sparse trace comprises a plurality of node identifiers in the second set of service interactions. Based on the node identifiers in the sparse trace and the node identifiers in the plurality of routes, the sparse trace is matched to one or more of the routes.

Interaction reconstruction in a service-oriented system

Methods, systems, and computer-readable media for implementing optimization of a service-oriented system using trace data are disclosed. One or more call graphs are determined based on trace data for a plurality of service interactions between individual ones of a plurality of services. The trace data comprises performance data for the service interactions. The call graphs comprise one or more call paths, and the call paths represent individual ones of the service interactions between the services. An optimized configuration for the services is determined based on the performance data. The optimized configuration improves a total performance metric for one or more call paths in the one or more call graphs. A location of one or more of the services is modified based on the optimized configuration.

Service-oriented system optimization using trace data

Methods and systems for implementing trace deduplication are disclosed. One or more interactions between a plurality of components are determined. The interactions relate to a trace. A fingerprint for the trace is generated. The fingerprint is compared with a plurality of stored fingerprints representing a plurality of stored traces. The trace is stored if the fingerprint does not match any of the plurality of stored fingerprints. The trace is determined to be a duplicate if the fingerprint matches one of the plurality of stored fingerprints.

Trace deduplication and aggregation in distributed systems

Methods and systems for implementing test coverage measurement are disclosed. A first set of interactions among a set of production services are determined. The first set of interactions comprises a plurality of service requests between individual ones of the set of production services. A second set of interactions among a set of test services are determined. The second set of interactions comprises a plurality of service requests between individual ones of the set of test services. A test coverage metric is generated. The test coverage metric indicates an amount of the first set of interactions that are covered by the second set of interactions.

Integrated test coverage measurement in distributed systems

Disclosed are various embodiments for generating and updating recommendations for merchants. A recommendation for a merchant in an electronic marketplace is evaluated to determine whether a corresponding action for the recommendation has been completed in response to a recommendation query for the merchant. A list of recommendations that correspond to actions that have not yet been completed for the merchant is generated.

Growing community of inventors

Bellevue, WA, United States of America

Anton Vladilenovich Goldberg