Cluster Workload Sharing via Community Manager Tokens
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Clusters of computing nodes often face resource constraints, leading to inability to handle high volumes of service requests, resulting in refused or failed requests due to insufficient resources.
Innovation Solution
A community of clusters shares workload by a community manager that monitors resource availability and utilizes tokens to establish secure borrower-lender relationships, allowing overloaded clusters to request and execute functions from resource-available clusters, thereby preventing refusal or failure of service requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a cluster handles high volumes of service requests, then productivity increases, but resource exhaustion occurs leading to request refusal or failure
Solution Approach 1:
The system segments the service request handling capability across multiple independent clusters. When one cluster experiences resource exhaustion, the workload is segmented and redistributed to other clusters in the community, allowing the system to maintain high productivity while preventing individual cluster failure from causing overall system failure.
Solution Approach 2:
Clusters in the community are designed with multi-functionality, where any cluster can potentially execute any function requested by the community. This universal capability allows flexible workload redistribution when resource constraints occur, ensuring that service requests can be fulfilled even when original requesting clusters are overloaded.
2Productivity
If clusters share workload through centralized scheduling, then resource utilization improves, but system complexity increases
Solution Approach 1:
Each cluster autonomously monitors its own resource status and makes decisions about accepting or rejecting service requests based on its current capacity. This self-service approach eliminates the need for complex centralized scheduling, as clusters independently manage their resources while participating in community-wide workload sharing through simple reject codes.
Solution Approach 2:
The community manager acts as a lightweight intermediary that maintains a registry of available functions across clusters but does not perform complex scheduling. It provides a simple interface for function discovery and routing, while the actual workload distribution emerges from individual cluster decisions, reducing overall system complexity.
3Adaptability or versatility
If clusters form a community to share resources, then adaptability increases, but security risks from unauthorized resource usage increase
Solution Approach 1:
Clusters perform preliminary authentication by verifying reject codes against a registry of valid codes maintained by the community manager. This preliminary validation ensures that only authorized clusters can request workload rejection and transfer, preventing unauthorized resource consumption while enabling legitimate workload sharing across the community.
Solution Approach 2:
The system implements feedback mechanisms where clusters report their resource status and receive feedback about available functions from the community manager. This continuous feedback loop enables clusters to adaptively adjust their workload acceptance decisions based on current community conditions, balancing adaptability with resource protection.
Data Source
AI summary
Example techniques for execution of functions by clusters of computing nodes are described. In an example, if a cluster does not have resources available for executing a function for handling a service request, the cluster may request another cluster for executing the function. A result of execution of the function may be received by the cluster and used for handling the service request.