Synthetic Alert Feeds for Accurate Data Center Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large-scale networked systems face challenges in ingesting and managing vast amounts of data from multiple infrastructure layers, with current monitoring technologies failing to accurately identify non-alert triggering violations, leading to alert fatigue and inaccurate analytics due to the exclusion of such violations.
Innovation Solution
A monitoring tool that generates two separate data feeds: one for alert triggering violations and another including synthetic alerts for non-alert triggering violations, providing a comprehensive view of a data center's performance by leveraging additional 'noise' to identify patterns and anomalies, thus aiding in predictive analytics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If alert trigger rule sets are applied to filter violations, then alert fatigue is reduced, but non-alert triggering violations are excluded leading to inaccurate analytics
Solution Approach 1:
The patent segments the data feed into two separate feeds: one containing only alert-triggering violations for operational response, and another containing all violations including non-alert triggering ones for comprehensive analytics. This segmentation allows each feed to serve its specific purpose without the harmful effects of mixing them.
Solution Approach 2:
The patent introduces an intermediary processing layer that generates synthetic alerts for non-alert triggering violations. These synthetic alerts act as a mediator between the raw violation data and the analytics system, enabling comprehensive analytics without triggering actual alert fatigue in the operational system.
2Loss of information
If all violations are included in the data feed, then analytics accuracy is improved, but alert fatigue increases due to excessive noise
Solution Approach 1:
The patent divides the violation data into two distinct segments: alert-triggering violations that require immediate operational response, and non-alert triggering violations that are included for analytics purposes. This segmentation prevents information loss while avoiding alert fatigue.
Solution Approach 2:
The patent creates a copy of the violation data feed that includes all violations with synthetic alerts generated for non-alert triggering ones. This copy is used for analytics without affecting the operational alert system, thus preserving information completeness while avoiding alert fatigue.
3Productivity
If synthetic alerts are generated for non-alert triggering violations, then predictive analytics are enhanced, but data processing complexity increases
Solution Approach 1:
The system automatically generates synthetic alerts for non-alert triggering violations based on predefined rules, eliminating the need for manual intervention. This self-service approach enhances predictive analytics while keeping the processing complexity manageable through automation.
Solution Approach 2:
The patent changes the parameter state of violations by generating synthetic alerts that transform non-alert triggering violations into alert-like structures. This parameter change enables comprehensive analytics without actually triggering operational alerts, balancing analytics enhancement with manageable complexity.
Data Source
AI summary
Systems and methods provide for automatically generating a data model that includes a first data feed conforming to industry standards where only alerts for alert triggering violations are provided. The data model further comprises a second data feed that includes both the alerts from the first data feed and a plurality of synthetic alerts for any violations that occur in a data center but do not qualify as alert triggering violations. This second data feed provides a complete picture of the performance of a data center's devices and allows for accurate analytics.