Subnet Manager Topology Graph for Middleware Fabric
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current middleware machine environments face challenges in maintaining a well-defined subnet topology and ensuring fault tolerance and scalability, particularly in InfiniBand Architecture systems, where complex state management and link failures can lead to inconsistent connectivity and reduced performance.
Innovation Solution
A system and method that employs a subnet manager to maintain a topology graph of connected ports and periodically reevaluate the states of fabric components, using subnet management agents to ensure correct responsiveness and adapt routing decisions, thereby maintaining a well-defined subnet topology and ensuring operational connectivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the subnet manager maintains a complete topology graph and periodically reevaluates states of all fabric components, then the reliability and fault tolerance are improved, but the complexity of state management and processing overhead increase
Solution Approach 1:
The patent segments the fabric components into different types (end devices, switches, initiators, targets) and maintains separate state information for each type. The topology graph is divided into multiple subgraphs representing different fabric domains, allowing independent management and evaluation of each segment rather than treating the entire fabric as a single complex entity.
Solution Approach 2:
The patent introduces topology templates as intermediary structures that define expected state configurations for different fabric component types. These templates act as mediators between the actual fabric state and the management system, providing a standardized framework for state evaluation and reducing the complexity of direct state management.
2Reliability
If the system performs comprehensive state reevaluation of all fabric components, then the connectivity consistency is improved, but the processing time and performance overhead increase
Solution Approach 1:
The patent implements periodic state reevaluation at predetermined intervals rather than continuous monitoring. The subnet manager schedules regular topology discovery and state verification operations, balancing connectivity consistency with processing time by performing evaluations at optimized intervals based on fabric stability and change detection.
Solution Approach 2:
The patent enables fabric components to self-report their state changes through asynchronous notifications to the subnet manager. When a fabric component detects a state change, it autonomously notifies the management system, which then performs targeted reevaluation only for affected components rather than comprehensive system-wide evaluation.
3Measurement precision
If the subnet manager monitors and responds to all fabric component states, then the fault detection capability is improved, but the system scalability is reduced
Solution Approach 1:
The patent applies different monitoring granularities and state evaluation criteria to different fabric component types and locations. End devices, switches, initiators, and targets each have customized state parameters and evaluation rules tailored to their specific functions. This localized approach enables precise fault detection for each component type while reducing the overall monitoring burden compared to uniform system-wide monitoring.
Solution Approach 2:
The patent introduces a hierarchical dimension to state management by organizing fabric components into multiple levels (fabric domain, switch level, port level). The topology graph represents relationships across these dimensions, allowing the system to scale by adding depth to the hierarchical structure rather than increasing horizontal monitoring complexity. This dimensional organization enables manageable fault detection in large-scale fabrics.
4Measurement precision
If the system maintains detailed topology information for all ports and connections, then the routing accuracy is improved, but the memory requirements and data management complexity increase
Solution Approach 1:
The patent creates simplified topological copies and abstractions of the physical fabric structure. Instead of storing complete detailed representations of all physical connections, the system maintains topological graphs that capture essential routing information in an abstracted format. These topological copies retain sufficient detail for accurate routing decisions while using significantly less storage than complete physical topology documentation.
Data Source
AI summary
A system and method can support a middleware machine environment. The middleware machine environment can include a subnet manager and a plurality of fabric components, wherein each said fabric component is associated with a subnet management agent (SMA). The subnet manager maintains a topology graph that indicates how ports are connected in the subnet. The subnet manager operates to periodically reevaluate one or more states associated with each fabric component in the subnet by determining whether each SMA is responding correctly to a request.