Veto-Based Telemetry System for Cloud Service Health
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions fail to effectively measure and address the overall user experience of cloud-based online services, as they focus on individual component reliability rather than the holistic performance of the service, making it difficult to identify and fix issues affecting the user's experience.
Innovation Solution
A veto-based system that collects and processes telemetry data across multiple components to create a unified metric for measuring the quality of experience, using session identifiers and vetoes to invalidate sessions that do not meet predefined performance metrics, providing a holistic view of product health and enabling rapid iteration on identified problems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If comprehensive reliability telemetry is collected from many system components, then the ability to detect anomalies is improved, but the difficulty of collecting and sorting telemetry signals to meaningfully assess user experience quality increases
Solution Approach 1:
The patent segments the complex telemetry data collection problem by introducing distinct functional modules: a telemetry data collection module that gathers signals from multiple components, a data processing module that filters and sorts the signals, and a user experience assessment module that generates meaningful metrics. This segmentation allows each module to handle specific tasks, reducing overall system complexity while maintaining comprehensive anomaly detection capability.
2Reliability
If individual component health is measured and improved, then component reliability is improved, but the ability to measure overall user experience quality deteriorates
Solution Approach 1:
The patent merges individual component telemetry signals into a unified user experience quality metric by collecting performance data from multiple components (storage, compute, network, database) and processing them together through a centralized data processing module. This merging approach enables the system to assess overall user experience quality rather than just individual component health, while still maintaining visibility into component-level reliability through the same infrastructure.
3Difficulty of detecting and measuring
If detailed telemetry data is collected from multiple components, then the ability to identify problem sources is improved, but the difficulty of coordinating across multiple product development teams increases
Solution Approach 1:
The patent introduces a centralized user experience assessment system as an intermediary that receives telemetry data from multiple product development teams' components, processes the data uniformly, and generates consolidated user experience quality metrics. This intermediary layer simplifies coordination by providing a single point of data aggregation and analysis, reducing the complexity of cross-team collaboration while maintaining detailed problem source identification capability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The performance of a cloud-based software product over time is determined by collecting telemetry data representing whether different features of online sessions of the software product are operating properly. The telemetry data represents shared performance metrics of the software product across different participants and components participating in an online session. The collected telemetry data is correlated with session identifiers identifying the online session from which the telemetry data was collected. The telemetry data for an online session is processed to establish a unit of failure when the telemetry data indicates that the online session operated outside of predefined performance metrics. The unit of failure is a function of vetoes applied to a candidate list of online sessions indicating that the online session may have problems. The performance of the software product may be determined as a function of the unit of failure over time.