Service metric analysis from structured logging schema of usage data

摘要

Technologies are generally described to provide a passive monitoring system employing a logging schema to track usage data in order to analyze performance and reliability of a service. The logging schema may be configured to track user requests as each request is received and processed at individual subsystems of the collaborative service. A logging entry may be created at a data store of the service, where the logging entry includes a subsystem name, an operation performed by the subsystem to fulfill the request, and start and end times of the operation. The logging schema may also detect errors fulfilling the requests, and may classify detected errors into a bucket, where each bucket denotes a failure scenario. Reliability of the service may be calculated based on analysis of the buckets to compute error rates. Reports may be generated to enable continuous monitoring of a performance and reliability of the system.

主权项

1. A method executed at least in part in a computing device to provide a passive monitoring system employing, a logging schema to track requests at a service, the method comprising:
detecting a request received by the service; creating a logging entry for the request at a data store associated with the service; detecting an error in processing the request at the service; recording the detected error with the logging entry for the request; classifying the detected error into an error bucket, wherein the error bucket denotes a failure scenario; and determining a reliability of the service based on a percentage of success buckets and error buckets, wherein synthetic requests are distinguished from real user requests and synthetic request data is removed to provide an accurate determination of the reliability of the service.