Autonomous Singleton Task Leasing Protocol
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In large-scale network services, efficiently processing singleton tasks across multiple server instances is complicated by the lack of global bootstrap concepts and minimal awareness between instances, leading to potential human intervention delays and errors when new servers are introduced or existing ones fail.
Innovation Solution
A leasing model protocol is implemented across multiple server instances, using a database to manage lease durations and facilitate optimistic concurrency, allowing a designated subset of server instances to process singleton tasks according to contracts, ensuring compliance and minimizing state management overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple discrete server instances are introduced to increase robustness and capacity, then system reliability and capacity improve, but task assignment complexity and state management overhead increase
Solution Approach 1:
The patent segments the task assignment problem into individual lease records stored in a database. Each singleton task is represented by a separate lease row that independently tracks expiration time and holder information. This segmentation allows each server instance to independently manage task assignments without complex global coordination, resolving the contradiction between system robustness and task assignment complexity.
Solution Approach 2:
The patent introduces a database as an intermediary between server instances and task management. The database stores lease information and provides a simplified interface for servers to claim and manage tasks through database operations (INSERT, UPDATE, SELECT). This intermediary eliminates the need for complex inter-server coordination protocols, reducing task assignment complexity while maintaining system reliability across multiple instances.
2Adaptability or versatility
If server instances operate with minimal interaction and state information, then system scalability improves, but task performance reliability deteriorates
Solution Approach 1:
The patent enables server instances to autonomously manage task assignments through self-service operations. Each server independently checks database leases, claims tasks by updating expiration times, and manages its own task execution without requiring interaction with other servers. This self-service model maintains scalability with minimal inter-server state while ensuring reliable task completion through individual server responsibility.
Solution Approach 2:
The patent implements preliminary action by pre-establishing lease records in the database before tasks are assigned. Each lease row contains pre-configured expiration times and holder information, allowing servers to immediately claim and execute tasks without prior coordination. This preliminary setup enables servers to operate independently while maintaining reliable task performance through predetermined lease contracts.
3Reliability
If human intervention is used for task assignment, then task performance reliability improves, but system response time and operational efficiency worsen
Solution Approach 1:
The patent replaces human intervention with automated self-service task assignment. Server instances autonomously check database leases, claim tasks by executing database UPDATE operations, and manage task execution without human involvement. This automation eliminates human response delays while maintaining reliable task performance through systematic lease-based assignment and expiration management.
Solution Approach 2:
The patent implements feedback mechanisms where servers continuously monitor database lease status and adjust their behavior accordingly. When a lease expires or becomes unavailable, servers receive feedback through database query results and automatically claim alternative tasks or wait for lease expiration. This automated feedback loop maintains task performance reliability while eliminating human intervention delays.
Data Source
AI summary
Large scale internet services may be implemented using multiple discrete server instances. Some tasks of the large scale internet services may be singleton tasks, which may be advantageously processed by a sub-set of the server instances (e.g., merely one instance). Accordingly, as provided herein, a singleton task may be processed in a reliable manner based upon one or more instances of a protocol executed across a set of arbitrary autonomous server instances. In one example, the protocol may determine whether a lease for a singleton task is valid or expired. If the lease is expired, then an attempt to claim the lease may be performed by updating a current lease expiration with a new lease expiration. If the attempt is successful, then the singleton task may be processed until the new lease expiration expires.