Resource Quality Estimation in Resource Sharing Platforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional resource sharing platforms face challenges in determining which resource offers to accept and reject, as they lack logic to evaluate the quality of resource offers and predict future offers, leading to suboptimal resource allocation due to unpredictable workload changes across slave machines.
Innovation Solution
A resource sharing system that estimates resource performance by measuring and storing data on slave machines, allowing frameworks to query historical and recent performance trends to make informed decisions between equivalent resource offers, incorporating a performance estimator task or application to augment resource offers with quality metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional frameworks accept resource offers based on basic availability checks, then resource allocation is simple and fast, but resource quality and throughput cannot be optimized
Solution Approach 1:
The patent introduces a performance service as an intermediary component that collects, stores, and provides performance data about slave machines. This mediator allows frameworks to access quality information without implementing complex collection and analysis logic themselves, thus improving resource allocation efficiency while keeping framework complexity manageable.
Solution Approach 2:
The system implements feedback mechanisms where performance data from slave machines is continuously collected and fed back to frameworks through the performance service. This feedback loop enables frameworks to make informed decisions about resource offers based on historical and current performance metrics, optimizing throughput without requiring frameworks to build complex monitoring systems.
2Measurement precision
If frameworks make resource decisions without performance data, then decision-making is fast and simple, but resource quality cannot be evaluated
Solution Approach 1:
The performance service pre-collects and stores performance data about slave machines before frameworks need to make resource decisions. By having this information readily available in advance, frameworks can quickly evaluate resource offers with accurate quality metrics without spending time collecting data during the decision-making process.
Solution Approach 2:
The system creates copies of performance data and makes them available through the performance service. Frameworks can query these pre-prepared data copies to evaluate resource offers quickly, avoiding the time cost of real-time measurement while still achieving precise quality evaluation.
3Adaptability or versatility
If all slave machines are treated equally in resource offers, then allocation logic is simple, but frameworks cannot distinguish between high-quality and low-quality resources
Solution Approach 1:
The patent applies local quality by attaching specific performance metrics to individual resource offers from specific slave machines. Instead of treating all resources uniformly, the system provides frameworks with quality information localized to each resource offer, enabling frameworks to make nuanced decisions about which specific resources to accept based on their performance characteristics.
Data Source
AI summary
Methods and apparatus for providing resource quality estimation in a resource sharing platform having a master and resources on slaves. Embodiments include storing resource performance data for the resources and estimating resource performance using the stored performance data. A framework can use the estimated resource performance data to select between a first resource offer and a second resource offer, which can be equivalent.