Capacity Attribution Engine for Shared Resource Pools
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In resource-on-demand systems like data centers, accurately determining the impact of applications on resource capacity is challenging due to varying demand patterns, leading to inefficient resource allocation and billing issues.
Innovation Solution
A system comprising a resource manager and a capacity manager that uses traces of resource utilization to determine required capacity and allocate resources effectively across workloads, incorporating a capacity attributed determination engine to calculate the portion of capacity attributed to each application, including unused capacity, using Classes of Service (CoS) constraints and probability-based resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If resources are allocated to accommodate peak demand of applications, then reliability of service is improved, but device complexity and resource waste increase
Solution Approach 1:
The patent implements dynamic resource allocation by continuously monitoring application performance metrics and adjusting resource allocation in real-time based on actual demand patterns. This allows the system to maintain service reliability during peak periods while automatically reducing allocations during low-utilization periods, avoiding the need for static over-provisioning that increases complexity.
Solution Approach 2:
The system changes allocation parameters dynamically by modifying resource allocation thresholds and thresholds based on observed application behavior patterns. By adjusting these parameters according to actual usage patterns rather than fixed peak demand assumptions, the system simplifies resource management while maintaining reliability.
2Reliability
If resources are allocated based on peak demand, then service reliability is improved, but resource utilization efficiency deteriorates
Solution Approach 1:
The patent employs dynamic resource allocation that adjusts allocations based on real-time monitoring of application performance and actual resource utilization patterns. This allows the system to maintain adequate resources for reliability during peak periods while automatically scaling back during low-utilization periods, thereby improving overall resource utilization efficiency.
Solution Approach 2:
The system enables applications to effectively self-regulate resource allocation by monitoring their own performance metrics and triggering resource adjustments based on predefined thresholds. This self-service mechanism improves resource utilization efficiency by ensuring resources are allocated based on actual need rather than conservative peak-demand provisioning.
3Productivity
If resources are shared among applications, then resource utilization efficiency is improved, but measurement precision of individual application impact deteriorates
Solution Approach 1:
The patent segments resource allocation and measurement by creating isolated monitoring contexts for each application. Even though resources are physically shared, the system virtually segments resource usage tracking by capturing and attributing resource consumption metrics to specific applications through unique identifiers and context propagation, enabling precise measurement of individual application impact while maintaining efficient shared resource utilization.
Solution Approach 2:
The patent introduces an intermediary measurement layer that sits between the shared resource pool and individual applications. This intermediary component captures, attributes, and tracks resource consumption for each application separately, enabling precise measurement of individual application impact while allowing efficient shared resource allocation. The intermediary maintains detailed attribution records without interfering with the shared resource utilization.
4Reliability
If resources are allocated to applications running at peak demand, then service reliability is improved, but loss of substance (resource waste) increases
Solution Approach 1:
The patent implements dynamic resource allocation that continuously monitors application performance metrics and adjusts resource allocations in real-time. During peak demand periods, the system maintains adequate resource allocations to ensure service reliability. During low-utilization periods, it automatically reduces allocations to minimize resource waste, eliminating the need for static over-provisioning.
Solution Approach 2:
The system dynamically changes resource allocation parameters based on observed application behavior patterns and actual utilization metrics. By adjusting allocation thresholds and parameters according to real-world usage rather than conservative peak-demand assumptions, the system maintains service reliability while minimizing resource waste through optimized parameter settings.
Data Source
AI summary
A required capacity for applications is determined. The applications are run on a shared pool of resources and each application belongs to at least one class of service of multiple classes of service. A portion of the required capacity attributed to each class of service is determined based on traces for the applications running on the shared pool of resources. A portion of the required capacity attributed to each application in each class of service is determined from the portion of the required capacity attributed to each class of service.