Intelligent Telemetry Collection via Workload Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing device management applications negatively impact remote device performance by initiating periodic collection of system state information during high workloads and are limited in the number of devices that can be simultaneously monitored, leading to delayed and outdated data collection.
Innovation Solution
The solution involves predicting future workloads and collection durations for each remote device using machine learning models, scheduling collections during idle times, and grouping them into chunks to minimize performance impact and ensure timely data gathering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If periodic collection of system state information is initiated at regular intervals, then timely data collection is improved, but remote device performance deteriorates due to high workload during collection periods
Solution Approach 1:
The system dynamically adjusts the timing of data collection based on the current workload state of remote devices. Instead of fixed periodic intervals, the scheduler monitors device workload metrics and adapts collection timing to execute during low-utilization periods, thereby maintaining data timeliness while minimizing performance impact
Solution Approach 2:
The system performs preliminary assessment of device workload before initiating data collection. By evaluating predicted workload metrics in advance, the scheduler proactively identifies suitable time windows for collection that avoid high-utilization periods, preventing performance degradation before it occurs
2Loss of time
If system state information is collected from all remote devices simultaneously, then data freshness is improved, but the management station becomes overwhelmed and collection capability deteriorates
Solution Approach 1:
The system segments the population of remote devices into multiple groups or batches. Instead of collecting data from all devices simultaneously, the scheduler distributes collection tasks across different time windows and resource pools, managing the complexity of monitoring thousands of devices through structured partitioning
Solution Approach 2:
The system introduces an additional scheduling dimension by organizing device collections into hierarchical groups and time-based batches. This multi-dimensional approach transforms the overwhelming simultaneous collection task into manageable segments across different organizational layers, enabling scalable monitoring of large device populations
3Productivity
If a fixed limited number of devices are monitored at a time, then management station resource utilization is improved, but data collection coverage deteriorates leading to outdated information
Solution Approach 1:
The system implements periodic rotation through multiple device groups, where each group is monitored intensively while others are temporarily excluded. This periodic cycling ensures all devices receive regular attention over time, maintaining comprehensive coverage while respecting management station resource constraints through structured temporal distribution
Data Source
AI summary
A system and method intelligently collect performance data from managed electronic devices. A machine learning model (e.g. linear time series forecasting) is used to predict a future workload for each of a selection of devices, and a regression analysis is used to predict how long is likely to be required to collect performance state from each component of each device. These data are then mapped together to predict future overall idle periods of each device, together with components whose performance data may be collected during those periods. The components are grouped in batches according to a relevance order that itself may be determined by applying a machine learning model such as k-nearest neighbors. Then, performance data are collected according to the batches. In this way, performance data may be collected in chunks while avoiding a negative impact on execution of the primary functions of the managed devices.