Network Service Demand Prediction Using Usage Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing providers face challenges in determining true demand for network services due to usage constraints, leading to inaccurate scaling and infrastructure provisioning based on truncated demand rather than actual demand.
Innovation Solution
A computing environment is configured to predict true demand by generating a demand curve that includes a truncated usage distribution and a predicted usage portion, using historical data and statistical methods to extrapolate usage beyond constraints, allowing for more accurate scaling and resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If usage constraints are imposed on network services to prevent unlimited use, then service stability and resource protection are improved, but the ability to determine true demand is worsened due to truncated usage data
Solution Approach 1:
The system performs preliminary actions by collecting usage data up to the constraint limit and storing it for later analysis. The demand curve generation is done in advance to predict future demand patterns, allowing the system to prepare for scaling decisions before actual demand exceeds constraints.
Solution Approach 2:
The patent introduces an intermediary demand curve generation system that acts as a mediator between the constrained usage data and the true demand determination. This intermediary system uses statistical methods and machine learning to bridge the gap between truncated observations and actual demand, allowing accurate measurement without violating service constraints.
2Object-affected harmful factors
If usage constraints cap the number of network calls, then resource overload is prevented, but infrastructure scaling decisions become inaccurate based on truncated demand data
Solution Approach 1:
The system implements feedback by continuously monitoring usage patterns up to the constraint limit and using this data to generate demand curves that predict true demand. This feedback loop enables accurate scaling decisions without requiring usage data beyond the constraint, allowing the system to optimize infrastructure provisioning based on predictive analytics rather than truncated observations.
Solution Approach 2:
The patent changes parameters by transforming constrained usage data into predictive demand curves using statistical transformations and machine learning models. This parameter transformation allows the system to derive meaningful demand information from constrained data, improving scaling efficiency without removing usage constraints.
3Ease of operation
If administrators monitor usage to determine demand, then service optimization is improved, but only truncated demand is visible due to usage limits
Solution Approach 1:
The patent replaces the mechanical approach of directly observing usage data with a computational approach using demand curve generation. Instead of relying on raw usage monitoring, the system uses statistical methods and machine learning algorithms to infer true demand from constrained data, substituting direct measurement with predictive modeling to recover lost information.
Solution Approach 2:
The system creates a composite demand curve that combines multiple elements: observed usage data up to the constraint, statistical distributions, and machine learning predictions. This composite structure allows the system to reconstruct the complete demand picture by integrating different data sources and methodologies, compensating for the information lost due to usage constraints.
Data Source
AI summary
Various embodiments are disclosed for predicting true demand of network services that impose usage constraints. A computing environment may be configured to provide a network service for a user account, where a usage constraint for the network service is imposed on the user account. A truncated usage distribution of the network service may be generated based on usage history data. A demand curve of the network service may be generated for the user account, where the demand curve comprises at least a portion of the truncated usage distribution and a predicted usage of the at least one network service beyond the usage constraint. One or more actions may be performed based on the demand curve, such as increasing the usage constraint.