Yield Management Framework for Computing Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing resource management techniques in computing centers only consider processor power and memory for client computing needs, separate from service definition and pricing, failing to optimize revenue and resource allocation effectively.
Innovation Solution
Implementing a yield management framework that integrates demand-side information and revenue changes through price and demand segmentation, using a management model to determine optimal resource allocation based on combinations of price levels and service levels, incorporating historical and predicted data to maximize revenue and achieve management goals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing resource management techniques are used that only consider processor power and memory, then resource allocation is simple, but revenue optimization is not achieved
Solution Approach 1:
The patent segments the resource management problem into multiple dimensions: resource types (processor, memory, storage), service levels, price levels, and client characteristics. This segmentation allows the system to evaluate different combinations systematically rather than treating resource allocation as a single-dimensional problem, thereby achieving revenue optimization while maintaining manageable complexity through structured analysis.
Solution Approach 2:
The patent introduces multiple parameters beyond traditional resource allocation (processor power and memory), including service level parameters, price parameters, and client-specific parameters. By changing the parameter set from simple binary allocation to multi-parameter optimization, the system achieves revenue optimization while the complexity is controlled through the use of a structured management model that evaluates parameter combinations systematically.
2Productivity
If resource allocation is separated from service definition and pricing, then each function is simple and independent, but overall system optimization is not achieved
Solution Approach 1:
The patent merges previously separate functions (resource allocation, service definition, and pricing) into a unified management model. This model simultaneously considers resource constraints, service level requirements, and price optimization, allowing the system to achieve overall optimization by evaluating the interplay between these functions rather than treating them as independent silos.
Solution Approach 2:
The management model serves multiple functions simultaneously: it allocates resources, defines service levels, sets prices, and optimizes revenue. This multi-functional approach achieves system-wide optimization while the complexity is managed through a single integrated model that handles all these functions coherently rather than requiring separate complex systems for each function.
3Productivity
If fixed revenue or cost is assumed for service satisfaction, then pricing is simple, but revenue maximization potential is not realized
Solution Approach 1:
The patent transitions from fixed revenue/cost assumptions to dynamic pricing and service level determination. The management model evaluates multiple price levels and service level combinations, selecting the optimal combination based on current resource availability, demand characteristics, and revenue optimization goals. This dynamic approach maximizes revenue potential while complexity is managed through systematic evaluation of discrete option combinations.
4Productivity
If only current client computing needs are considered, then resource allocation is straightforward, but future demand and revenue opportunities are not captured
Solution Approach 1:
The patent incorporates demand prediction and historical data analysis into the resource management process, allowing the system to anticipate future demand patterns and prepare optimal resource allocation strategies in advance. The management model uses historical data and predicted data to forecast demand, enabling proactive revenue optimization rather than merely reacting to current requests, while complexity is managed through structured data processing pipelines.
Data Source
AI summary
Yield management techniques are provided. In one aspect of the invention, a technique for managing one or more computing resources comprises the following steps/operations. Data associated with at least one potential demand for use of the one or more computing resources is obtained. Then, a management model (e.g., a yield management model or a revenue management model) is generated in accordance with at least a portion of the obtained data, wherein the management model is operative to determine an allocation of the one or more computing resources based on combinations of price levels and service levels that may be offered to one or more users of the one or more computing resources so as to attempt to satisfy at least one management goal.