Workload Inspection and Rescheduling for Excessive Resource Usage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed computing environments face inefficiencies due to excessive resource usage by users, impacting other users and overall platform efficiency, with existing solutions failing to proactively address this issue beyond user activity.
Innovation Solution
A system that incorporates inspectors to analyze workload code and account information, determining a likelihood of excessive resource usage through pattern recognition and machine learning, and reschedules workloads based on a threshold to mitigate resource excess.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the distributed computing environment allows users to access a shared pool of configurable computing resources, then resource provisioning flexibility and user convenience are improved, but excessive resource usage by individual users can occur, impacting other users and overall platform efficiency
Solution Approach 1:
The system performs preliminary analysis of workload code and account information before executing workloads to predict potential excessive resource usage. By proactively identifying risky workloads through code pattern recognition and machine learning models, the system can prevent resource exhaustion before it occurs, maintaining platform efficiency while preserving resource provisioning flexibility for legitimate workloads.
Solution Approach 2:
The system implements a feedback mechanism where resource usage patterns are continuously monitored and fed back into the machine learning models. This enables dynamic adjustment of resource allocation decisions, allowing the system to adapt to emerging threats while maintaining efficient resource utilization. The feedback loop helps distinguish between legitimate high-resource workloads and malicious ones, resolving the contradiction between flexibility and efficiency.
2Reliability
If existing solutions monitor user activity to detect excessive resource usage, then some level of resource abuse detection is achieved, but the solutions fail to proactively address the issue by analyzing beyond just user activity
Solution Approach 1:
The system performs preliminary analysis of workload code and account information before executing workloads to predict potential excessive resource usage. By proactively identifying risky workloads through code pattern recognition and machine learning models, the system can prevent resource exhaustion before it occurs, maintaining platform efficiency while preserving resource provisioning flexibility.
Solution Approach 2:
The system introduces an intermediary inspection layer that analyzes workload code, account information, and resource usage patterns. This intermediary component bridges the gap between passive monitoring and active prevention by using machine learning models to predict excessive resource usage before it occurs, enhancing detection reliability while making proactive detection feasible.
3Productivity
If the system reschedules workloads based on predicted excessive resource usage, then platform efficiency is improved and impact on other users is reduced, but additional processing overhead is introduced for analyzing code and account information
Solution Approach 1:
The system applies different levels of inspection intensity to different workloads based on their risk profiles. High-risk workloads identified through code patterns or account information receive more rigorous analysis, while low-risk workloads undergo minimal inspection. This localized approach maintains platform efficiency by focusing computational resources on problematic workloads rather than uniformly processing all workloads, thereby reducing overall processing overhead.
Solution Approach 2:
The system dynamically adjusts inspection parameters such as analysis depth, model complexity, and sampling rates based on workload characteristics and current system conditions. This allows the system to optimize the balance between detection accuracy and processing overhead, maintaining high platform efficiency while adapting computational resource consumption to actual needs.
Data Source
AI summary
A method, computer program product and computer system for predicting excessive resource usage in a distributed computing environment is provided. A processor retrieves a portion of code associated with a workload in a distributed computing environment. A processor retrieves account information associated with the workload. A processor determines a likelihood that the workload is indicative of excessive resource usage based on the portion of code and the account information associated with the workload. In response to the likelihood of excessive resource usage exceeding a threshold, a processor reschedules the workload in the distributed computing environment.