Capacity Unit Prediction for Accurate Cloud Cluster Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing predictive models for cloud resource allocation are inaccurate, leading to either resource under-provisioning, resulting in poor service performance, or over-provisioning, causing resource waste, due to the dynamic nature of workload and the reliance on single-indicator heuristics.
Innovation Solution
A machine learning model combining wide and deep learning techniques is used to predict future resource usage by integrating discrete and time-series data, allowing for accurate capacity unit predictions, which inform cluster allocation decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If existing predictive models are used for resource allocation, then resource allocation decisions can be made automatically, but the prediction accuracy is low resulting in either resource under-provisioning or over-provisioning
Solution Approach 1:
The patent segments the resource allocation problem into multiple independent predictive models, each specialized for different workload types (CPU-intensive, memory-intensive, network-intensive, I/O-intensive). This segmentation allows each model to focus on specific patterns, improving overall prediction accuracy while maintaining automatic allocation through the ensemble of specialized models.
Solution Approach 2:
The system dynamically adapts to changing workload conditions by continuously monitoring resource usage patterns and adjusting predictions in real-time. The predictive models are updated based on historical data and current system state, enabling accurate automatic allocation even as workload characteristics evolve over time.
2Reliability
If resource allocation is increased to ensure service performance, then service quality improves, but resource waste increases due to over-provisioning
Solution Approach 1:
The system performs preliminary resource allocation based on accurate predictive modeling of future workload demands. By predicting resource needs in advance with high accuracy, the system allocates resources proactively to prevent service performance degradation while avoiding excessive allocation that would lead to waste. The preliminary action is refined through continuous monitoring and model updates.
Solution Approach 2:
The patent implements feedback mechanisms where actual resource usage and service performance metrics are continuously monitored and fed back to the predictive models. This feedback loop allows the system to learn from past allocations, refine predictions, and optimize the balance between ensuring service performance and minimizing resource waste over time.
3Loss of energy
If resource allocation is decreased to reduce waste, then resource efficiency improves, but service performance deteriorates due to under-provisioning
Solution Approach 1:
The system performs preliminary resource allocation based on accurate predictive modeling of future workload demands. By predicting resource needs in advance with high accuracy, the system allocates resources proactively to prevent service performance degradation while avoiding excessive allocation that would lead to waste. The preliminary action is refined through continuous monitoring and model updates.
4Device complexity
If single-indicator heuristics are used for capacity prediction, then the system complexity is reduced, but prediction accuracy deteriorates due to inability to capture dynamic workload patterns
Solution Approach 1:
The patent segments the resource allocation problem into multiple independent predictive models, each specialized for different workload types (CPU-intensive, memory-intensive, network-intensive, I/O-intensive). This segmentation allows each model to focus on specific patterns, improving overall prediction accuracy while maintaining automatic allocation through the ensemble of specialized models.
Solution Approach 2:
The system employs a universal framework that handles multiple workload types and resource dimensions through a consistent multi-factor predictive approach. The same underlying methodology is applied across different resource types (CPU, memory, network, I/O), providing accurate predictions for diverse workloads without requiring completely separate systems for each case.
Data Source
AI summary
Real-time workload for an application is converted into a capacity unit value. A trained machine-learning model receives the capacity unit data as input and generates a prediction of capacity usage for the future. Based on the prediction, additional clusters may be allocated for the application. A dataset for use in predicting future capacity unit usage by an application may be classified into two categories. A first category comprises time series data. A second category comprises service types, target user types, and other attributes with discrete characteristics. Discrete attributes are incorporated into a wide section and time-series data is integrated into a deep section. A wide and deep model combines results from the wide section and the deep section to generate a prediction of capacity units used by the application in future time periods. In response, an allocation system allocates a corresponding number of clusters.