Datacenter Monitoring via Time-Series Search and Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current datacenter management systems struggle to effectively monitor and visualize the complex configurations of physical and virtual entities within modern datacenters, as they often fail to understand new architectures and relationships between these entities, leading to incomplete monitoring and troubleshooting capabilities.
Innovation Solution
A management system utilizing time-series based modeling and natural language search engines to capture the evolving state of datacenters, including physical, virtual, and logical entities, and their relationships, enabling efficient data collection, visualization, and collaboration among users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing monitoring systems are used to track datacenter entities, then monitoring coverage is limited to a subset of components, but the system cannot effectively monitor or visualize the complete physical and virtual entity configurations
Solution Approach 1:
The patent implements a universal data model that can represent multiple types of datacenter entities (physical devices, virtual machines, logical networks) and their relationships in a unified framework. This allows the system to monitor diverse components across different layers (network, compute, storage) through a single adaptable model structure, resolving the contradiction between comprehensive monitoring coverage and adaptability to new architectures.
Solution Approach 2:
The system employs dynamic data models that can adapt to changing datacenter configurations and emerging entity types. The model structure allows for adding new entity types and relationships without requiring fundamental system changes, enabling the monitoring system to evolve with new datacenter architectures while maintaining comprehensive coverage.
2Reliability
If comprehensive data collection is implemented for all entities, then complete monitoring capability is achieved, but system complexity increases significantly
Solution Approach 1:
The patent segments the complex datacenter monitoring task into hierarchical layers (physical devices, virtual entities, logical relationships) represented in the data model. Each entity type has its own structured representation with specific attributes, allowing comprehensive data collection to be organized into manageable segments that can be processed and visualized systematically, reducing overall system complexity.
Solution Approach 2:
The unified data model acts as an intermediary layer between diverse datacenter entities and the monitoring system. This intermediate representation standardizes different entity types and relationships, simplifying the complexity of directly managing heterogeneous data sources while maintaining complete monitoring capability.
3Ease of repair
If detailed entity relationships are tracked, then troubleshooting capability is enhanced, but data processing and visualization become more difficult
Solution Approach 1:
The data model applies local quality by defining specific attributes and relationship types for different entity contexts. Each entity type (physical device, virtual machine, network component) has tailored properties relevant to its troubleshooting needs, allowing detailed relationship tracking to be presented in context-appropriate formats that simplify visualization and operation.
Data Source
AI summary
A datacenter management system uses data collection proxies to collect performance data and configuration data for different physical and virtual entities in the datacenter. A schema is used to represent the different entities, entity relationships, and entity properties in the datacenter. A search engine identifies the intent of a natural language based search query based on the schema and a datacenter dictionary. The search engine then searches the data based on the search query intent. A dictionary manager converts both periodic and aperiodic data into a time series. This allows the search engine to operate as a time machine identifying both performance data and configuration data for any selectable time period.