Consistent Hash Mapping for Cluster Resource Borrowing Conflicts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing resource allocation methods in distributed systems face borrowing conflicts when multiple busy clusters concurrently request resources from the same idle cluster, leading to increased delay and inefficiency.
Innovation Solution
A resource allocation method using a consistent hash algorithm to establish unique associations between clusters, allowing only a specific cluster to borrow resources from another, thereby eliminating concurrent borrowing conflicts and reducing delay overheads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple busy clusters concurrently borrow resources from the same idle cluster using random selection, then resource borrowing flexibility is improved, but borrowing conflicts increase and system efficiency deteriorates
Solution Approach 1:
The patent segments the resource borrowing process by dividing clusters into idle clusters and busy clusters, with each busy cluster pre-assigned to a specific idle cluster through hash ring mapping. This segmentation eliminates random selection and concurrent borrowing conflicts while maintaining resource borrowing flexibility.
Solution Approach 2:
The patent implements preliminary action by pre-establishing borrowing relationships between busy clusters and idle clusters using consistent hash algorithms before actual resource borrowing occurs. This pre-mapping prevents borrowing conflicts by ensuring each busy cluster has a predetermined target idle cluster to borrow from.
2Adaptability or versatility
If random resource query requests are sent to select idle clusters, then resource allocation adaptability is improved, but delay overheads increase
Solution Approach 1:
The patent uses consistent hash algorithms to pre-establish borrowing relationships between busy clusters and idle clusters before resource borrowing is needed. This preliminary mapping eliminates the need for random resource query requests and round-trip communications, significantly reducing delay overheads while maintaining allocation adaptability.
3Productivity
If concurrent borrowing requests are allowed from multiple busy clusters to the same idle cluster, then resource utilization is improved, but borrowing conflicts occur
Solution Approach 1:
The patent segments the system into idle clusters and busy clusters with dedicated borrowing relationships established through hash ring mapping. Each busy cluster is assigned to a specific idle cluster, preventing multiple busy clusters from concurrently borrowing from the same idle cluster and eliminating borrowing conflicts while maintaining high resource utilization.
Solution Approach 2:
The patent introduces a hash ring as an intermediary mechanism that mediates resource borrowing relationships. The hash ring pre-determines which idle cluster serves which busy cluster, acting as a mediator that prevents direct conflicts between multiple busy clusters seeking resources from the same idle cluster.
Data Source
AI summary
Embodiments of this application provide example resource allocation methods. One example method includes determining, by a first node, a hash value corresponding to a first cluster. The first node can obtain a hash ring, where the hash ring includes at least one hash value interval, and each hash value interval corresponds to one second cluster. The first node can determine a target second cluster based on the hash value corresponding to the first cluster, where the hash value corresponding to the first cluster belongs to a target hash interval, the target hash interval is a hash value interval in the at least one hash value interval, and the target hash interval corresponds to the target second cluster. The first node can establish an association relationship between the first cluster and a target second cluster, where the target second cluster is allowed to apply for a resource from the first cluster.