Dynamic Tree Overlay for Self-Organizing Grid Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computational grids face challenges in automatically discovering efficient clusters within self-organizing grids (SOGs) for effective scheduling of resources, as current methods require significant administrative overhead and are not scalable for large-scale systems.
Innovation Solution
A Minimum-delay Dynamic Tree (MDTree) overlay structure is used to organize nodes based on link delay, allowing for proactive and on-demand resource clustering with variable sizes, and a Group-Scheduling framework for efficient resource allocation and monitoring, incorporating distributed authorization and adaptive monitoring mechanisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a brute force approach is used to discover ad hoc clusters by periodically testing network performance characteristics between all pairs of resources, then complete cluster information can be obtained, but the system becomes infeasible for large scale systems due to excessive computational overhead
Solution Approach 1:
The patent segments the cluster discovery process into two distinct phases: an offline phase that pre-computes and stores cluster information using brute force methods, and an online phase that efficiently retrieves pre-computed cluster data. This segmentation allows the system to obtain complete cluster information without incurring excessive computational overhead during runtime, making large-scale systems feasible.
Solution Approach 2:
The patent performs cluster discovery computations in advance during an offline phase, storing the results for later use. By pre-computing network performance characteristics and cluster relationships before they are needed, the system avoids the need to perform expensive brute force calculations during online operation, thereby reducing real-time computational overhead while maintaining complete cluster information accuracy.
2Reliability
If manual methods are used to add and manage resources in structured grids, then resource organization can be controlled, but significant administrative overhead is required making it infeasible for individual resources
Solution Approach 1:
The patent implements self-service mechanisms where resources automatically discover and join clusters without manual administrative intervention. The system autonomously performs resource discovery, cluster formation, and management tasks, eliminating the need for significant administrative overhead while maintaining reliable resource organization. This enables individual resources to participate easily in the grid infrastructure.
Solution Approach 2:
The patent changes the operational parameters of resource management from manual, structured procedures to automated, dynamic processes. By transforming resource addition and management from human-operated tasks to system-autonomous operations, the patent maintains control and reliability while dramatically improving ease of operation and enabling widespread individual resource participation.
3Stability of the object's composition
If existing grid methods are used to organize resources in a structured way, then resource management can be controlled, but the administrative overhead becomes a significant barrier to participation
Solution Approach 1:
The patent introduces dynamic, adaptive cluster discovery and formation mechanisms that automatically adjust to changing grid conditions without requiring structured, manual resource organization. The system dynamically discovers clusters based on actual network performance characteristics and resource availability, maintaining structural stability through self-organization while eliminating the need for complex administrative overhead. This enables flexible participation of individual resources without compromising grid stability.
Data Source
AI summary
A cluster of nodes, comprising: a plurality of nodes, each having a security policy, and being associated task processing resources; a registration agent configured to register a node and issue a node certificate to the respective node; a communication network configured to communicate certificates to authorize access to computing resources, in accordance with the respective security policy; and a processor configured to automatically dynamically partition the plurality of nodes into subnets, based on at least a distance function of at least one node characteristic, each subnet designating a communication node for communicating control information and task data with other communication nodes, and to communicate control information between each node within the subnet and the communication node of the other subnets.