Arbitrator Mediated Lease Failure Detection in Distributed Nodes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing failure detection schemes in distributed computing systems can cause nodes to terminate prematurely or unnecessarily due to network congestion or failure, leading to a reduction in processing and storage resources.
Innovation Solution
Implementing delay and neutral arbitration results in lease protocols to allow both nodes to continue operating by utilizing a waiting period for arbitration requests, preventing unnecessary termination and allowing time for requests to reach the arbitrator.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional lease protocols are used for failure detection, then node failures can be detected, but nodes may terminate prematurely or unnecessarily due to network congestion or failure
Solution Approach 1:
The system performs preliminary actions by sending arbitration requests before making termination decisions. When a node detects lease failure, it sends an arbitration request to the arbitrator before terminating, allowing the arbitrator to assess the situation and provide arbitration results that prevent premature termination due to transient network issues.
Solution Approach 2:
An arbitrator node is introduced as an intermediary between monitoring nodes and subject nodes. The arbitrator receives arbitration requests from nodes experiencing lease failures and provides arbitration results that mediate between conflicting termination decisions, preventing unnecessary node terminations while maintaining reliable failure detection.
2Speed
If arbitration requests are sent immediately upon lease loss, then failure detection is rapid, but nodes terminate before arbitration can prevent unnecessary termination
Solution Approach 1:
The system sends arbitration requests as a preliminary action immediately upon detecting lease loss, enabling rapid failure detection. The arbitration request is sent before any termination decision is made, allowing the arbitrator to intervene and provide accurate termination decisions based on the arbitration result rather than premature local decisions.
3Reliability
If nodes terminate upon lease loss to maintain system integrity, then system reliability is maintained, but processing and storage resources are reduced
Solution Approach 1:
The arbitrator serves as an intermediary that receives arbitration requests from nodes experiencing lease failures and provides arbitration results that determine whether termination is necessary. This mediation mechanism maintains system integrity by preventing false terminations due to transient network issues while preserving resource capacity by allowing nodes to continue operating when appropriate.
Solution Approach 2:
The system implements feedback through arbitration results that provide information to nodes about whether termination should occur. The arbitrator analyzes arbitration requests and provides feedback in the form of arbitration results that guide node termination decisions, ensuring that terminations occur only when necessary for system integrity while preserving resources when possible.
Data Source
AI summary
Failure detection techniques in distributed computing systems are disclosed herein. In one embodiment, a method includes receiving an arbitration request from a first node in a computing system having a plurality of nodes interconnected by a computer network. Each of the nodes has a logic relationship with another node in the computing system. The arbitration request indicates that the first node is unable to establish a lease with a second node for a predetermined threshold period. The second node is logically related to the first node according to the logic relationship and is a default monitor for the first node for the lease. The method also includes in response to receiving the arbitration request from the first node, providing a neutral arbitration result to the first node, and allowing the first node to continue to operate without causing the second node to terminate itself.