Multi-layer Analytics Stack for Cloud Resource Forecasting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems face inefficiencies due to unavailability of allocated resources, even when they are idle, leading to underutilization and increased costs, as resources allocated to one project remain unavailable for others despite their idle state.
Innovation Solution
A multi-layer analytics service stack that predicts future resource utilization by analyzing historical data, allowing for accurate resource allocation and reclamation of dormant resources, thereby optimizing resource utilization and reducing costs through the use of a data lake interaction layer, transformation layer, data treatment layer, data partitioning layer, multi-forecasting engine layer, comparative evaluation layer, and presentation layer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cloud computing resources are allocated to a first computing project, then the resource is reserved for that project, but the resource remains unavailable for reallocation to a second computing project even when idle
Solution Approach 1:
The system performs preliminary actions by continuously monitoring resource utilization metrics before resources become completely idle, triggering reallocation opportunities in advance. The resource allocation system proactively identifies underutilized resources and initiates reallocation processes before the resources are fully released, ensuring both project reliability and improved utilization.
Solution Approach 2:
The patent implements dynamic resource allocation where resource assignments are not static but continuously adjusted based on real-time utilization monitoring. The system dynamically transitions resources between projects based on demand, allowing the same resource to serve multiple projects sequentially rather than being permanently bound to one project, thereby resolving the contradiction between reliability and productivity.
2Measurement precision
If multiple analytics models are evaluated to predict resource utilization, then prediction accuracy improves, but processing time and computational complexity increase
Solution Approach 1:
The system applies partial action by evaluating a subset of analytics models rather than all available models. It selectively applies modeling techniques based on the specific context and data characteristics, using simpler models when sufficient and more complex models only when necessary, thereby achieving good prediction accuracy without excessive processing time.
Solution Approach 2:
The patent segments the analytics modeling process into multiple independent evaluation stages. Different analytics models are evaluated separately and their results combined or compared, allowing the system to leverage multiple models for improved accuracy while managing computational complexity through structured segmentation of the evaluation process.
Data Source
AI summary
A multi-layer analytics service stack may generate forecasted utilization data. An input layer of the analytics service stack may receive input and designate cloud computing utilization data for analysis. A transformation layer of the analytics service stack may perform format transformations on the cloud computing utilization data, and the data may be prepared for analysis at a data treatment layer of the cloud computing utilization data. The treated and transformed cloud computing utilization data may be analyzed using multiple analytics models by a multi-forecasting layer of analytics service stack to generate the forecasted utilization data.