Cloud Resource Mapping for Mixed Graphics Workloads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing environments face inefficiencies in mapping applications to hardware, leading to suboptimal resource utilization and increased costs due to initial mappings not reflecting actual performance and utilization data, resulting in potential over or under utilization of hardware resources.
Innovation Solution
A method is introduced to generate a lower-cost mapping by determining baseline actual hardware performance utilization, calculating equivalent costs for each hardware class, and remapping applications to the hardware class with the lowest equivalent cost, while maintaining performance requirements, using performance monitoring data and conversion factors to adjust mappings dynamically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If applications are mapped to hardware using initial estimates, then the mapping process is simple and quick, but resource utilization is suboptimal and costs increase
Solution Approach 1:
The system performs preliminary actions by initially mapping applications to hardware based on estimates, then later remapping based on actual performance data. This allows the system to start with a simple initial configuration and optimize it subsequently, resolving the contradiction between initial simplicity and eventual optimization.
Solution Approach 2:
The system implements feedback mechanisms by monitoring actual hardware performance utilization and using this data to remap applications. The feedback loop collects performance data, analyzes it, and adjusts mappings accordingly, improving resource utilization while managing complexity through automated processes.
2Reliability
If hardware resources are allocated based on estimated performance, then allocation is straightforward, but over or under utilization occurs leading to increased costs
Solution Approach 1:
The system transitions from static initial mappings to dynamic remappings based on actual performance data. Applications are continuously monitored and reassigned to hardware that best matches their actual resource needs, preventing both over-utilization (waste) and under-utilization (inefficiency), thus maintaining reliability while reducing resource waste.
Solution Approach 2:
The system changes mapping parameters by switching from estimated performance parameters to actual measured performance parameters. This parameter change enables more accurate matching of applications to hardware, ensuring service-level agreements are met while minimizing resource waste and associated costs.
3Ease of operation
If applications are remapped based on actual performance data, then cost-effectiveness improves, but system complexity and remapping overhead increase
Solution Approach 1:
The system performs self-service by automatically monitoring its own performance, analyzing the data, and remapping applications without external intervention. This automation improves operational efficiency while managing complexity through self-managed processes, reducing the need for manual intervention despite increased system sophistication.
4Productivity
If initial mappings are used without adjustment, then system operation is simple, but resource allocation is suboptimal leading to higher costs
Solution Approach 1:
The system maintains continuous useful action by constantly monitoring performance and continuously optimizing mappings. Rather than periodic adjustments, the continuous monitoring and remapping process ensures hardware resources are always optimally utilized, improving productivity while the automated nature minimizes time loss through efficient data collection and analysis.
Data Source
AI summary
In one embodiment, a method includes, by a cloud services system, determining a baseline actual hardware performance utilization of a plurality of hardware computing devices for a plurality of applications in accordance with an initial mapping that maps a plurality of hardware resource classes to the plurality of applications, where each of the hardware computing devices is associated with one of the plurality of hardware resource classes, determining a lower-cost configuration in which each application is assigned to the hardware class having a lowest equivalent cost for that application, and, when the sum of the lowest equivalent costs for each application assigned to a particular hardware class by the lower-cost configuration is less than a threshold value, moving one or more applications from their initially-assigned hardware classes in the initial mapping to the particular hardware class.