Telemetry Data Segmentation for Low-Latency Platform Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing infrastructure faces challenges in providing low-latency and high-throughput platform telemetry data due to the overhead from verbose, context-rich data formats like CIM and Redfish, which increase latency and reduce scalability.
Innovation Solution
The solution involves separating telemetry data from context information and using telemetry management controllers and engines to deliver data via a fast path with minimal overhead, utilizing more compact formats and bypassing processing by the telemetry management controller, while maintaining compatibility with existing ecosystems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If telemetry data is provided using verbose context-rich formats like CIM and Redfish, then data completeness and context information are improved, but network data transfer overhead and processing latency increase
Solution Approach 1:
The patent segments telemetry data into two distinct paths: a fast path that transmits only essential telemetry values without context information, and a slow path that provides complete context-rich data. This segmentation allows consumers to receive critical data quickly while optionally obtaining detailed context information separately, thereby reducing latency for time-sensitive applications while preserving data completeness for comprehensive analysis.
Solution Approach 2:
The patent extracts and separates context information from telemetry data values. The fast path extracts and transmits only the essential numerical telemetry values (e.g., temperature readings, CPU utilization percentages), while the slow path handles the verbose context information (e.g., component identifiers, metadata). This extraction enables the fast path to achieve low-latency transmission by eliminating redundant contextual data.
2Loss of information
If telemetry data is provided using verbose context-rich formats like CIM and Redfish, then data completeness and context information are improved, but scalability is reduced
Solution Approach 1:
By segmenting the data delivery into fast and slow paths, the system can scale to handle larger volumes of telemetry data from multiple components without proportionally increasing network overhead. The fast path scales efficiently by transmitting only compact numerical values, while the slow path handles context information on-demand, allowing the system to accommodate growing infrastructure requirements.
Solution Approach 2:
The patent changes the data format parameter from verbose context-rich representations to compact numerical values for the fast path. This parameter change reduces the amount of data transmitted and processed, directly improving scalability. The system can now handle higher data volumes from additional components and sensors without linearly increasing network bandwidth requirements or processing loads.
3Measurement precision
If telemetry management controller processes all telemetry data, then data accuracy and context information are improved, but processing overhead and latency increase
Solution Approach 1:
The patent segments the processing workload by routing different types of data through different paths. The fast path bypasses the telemetry management controller for time-critical telemetry values, allowing these measurements to be transmitted with minimal processing delay. The slow path routes context-rich data through the controller for accurate processing and validation. This segmentation enables the system to maintain data accuracy for comprehensive analysis while achieving high processing throughput for real-time monitoring.
Solution Approach 2:
The fast path performs preliminary action by transmitting essential telemetry values directly to consumers before the telemetry management controller completes full processing. This preliminary transmission of critical data allows consumers to immediately react to time-sensitive events (e.g., hardware failures, performance degradation) while the controller continues processing detailed context information in the background.
Data Source
AI summary
In one embodiment, a request for telemetry data measured by a plurality of components of a computing platform is received from a computing device. Contextual information associated with the requested telemetry data is provided in a first communication, wherein the contextual information comprises information describing the plurality of components. An instance of the requested telemetry data is provided to the computing device, wherein the telemetry data is provided in a second communication that omits at least a portion of the contextual information describing the plurality of components.