Cloud Infrastructure Recommendation Using Usage Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cloud infrastructure recommendation systems do not effectively account for how applications use cloud hardware resources, leading to suboptimal performance and energy consumption, as they rely on predefined user requirements without considering actual usage patterns.
Innovation Solution
A method and apparatus that identify preferred cloud infrastructure systems by analyzing usage profiles, including performance and energy consumption data, to select systems with similar resource requirements, predicting energy consumption and performance, and recommending the most efficient options between private and public clouds based on these metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If cloud brokers base recommendations on predefined user requirements and template evaluations, then the recommendation process is simplified and faster, but the recommendations do not accurately reflect actual application usage patterns and resource consumption
Solution Approach 1:
The system performs preliminary actions by collecting and storing usage profiles from multiple applications before a recommendation is needed. These profiles include actual resource consumption data, performance metrics, and energy usage patterns. When a recommendation is requested, the system can quickly retrieve and compare pre-analyzed usage profiles rather than analyzing applications in real-time, thus maintaining fast response while ensuring accuracy based on actual usage data.
Solution Approach 2:
The system creates copies of usage profiles from executed applications and stores them in a database. These copied profiles can be reused for multiple recommendation queries without re-executing the applications, enabling fast recommendations while preserving the accuracy of actual usage measurements. The system matches new applications against these stored profile copies to generate recommendations.
2Measurement precision
If the system analyzes actual application execution to generate usage profiles, then the recommendations are more accurate regarding resource consumption, but the system complexity and data collection overhead increase
Solution Approach 1:
The system introduces an intermediary usage profile database that sits between the application executions and the recommendation process. This database stores pre-collected usage profiles including resource consumption, performance, and energy data. The intermediary layer simplifies the overall system by centralizing data collection and management, making the recommendation engine less complex as it can query pre-processed profiles rather than directly managing execution and analysis of multiple applications.
Solution Approach 2:
The system implements self-service by automatically executing applications, collecting their usage profiles, and storing them without manual intervention. The cloud broker automatically queries this self-maintained database when recommendations are needed, reducing the complexity of manual data collection and profile management while maintaining high measurement precision through automated actual execution data.
3Reliability
If the system collects and stores usage profiles from multiple applications, then better recommendations can be made, but the data storage and processing requirements increase
Solution Approach 1:
The system extracts and stores only the essential usage profile data needed for recommendations, such as resource consumption patterns, performance metrics, and energy usage characteristics. By taking out only the relevant features from complete application execution data, the system reduces the volume of stored information while maintaining the quality and reliability of recommendations. This selective extraction prevents data overload while preserving recommendation accuracy.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
A cloud infrastructure recommendation method to select a preferred private or public cloud system to execute a computer application comprises determining which applications in a register of computer applications, for which usage profiles comprising quantitative performance and energy consumption data are stored, have usage profiles which are similar to a usage profile of the application. From computing systems associated with the applications determined to have similar usage profiles, a first set of computing systems are selected which meet a minimum performance requirement or an energy consumption criterion. A second set of computing systems on the public cloud system which have the most similar resources to the computing systems of the first set are also selected. The cloud system which corresponds to the computing system having a lowest predicted energy consumption of those computing systems of the first and second sets that have a predicted performance meeting a desired performance level is identified as the preferred cloud system.