Dynamic Capacity Ceilings for Partitioned Data Processing Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for managing resource allocation in logically partitioned servers struggle to ensure optimal resource utilization under varying workloads, often leading to oversizing and increased complexity, power consumption, and performance degradation due to inflexible resource management policies.
Innovation Solution
A method for dynamically controlling processor resource access in a partitioned data processing system by recalculating and applying variable capacity ceilings (DC_Pi(t)) based on instantaneous resource consumption, workload measurements, ceiling states, and user-defined parameters, optimizing resource sharing and allocation across multiple partitions and servers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If fixed capacity ceilings are imposed on logic partitions to control resource consumption, then resource management compliance is improved, but system flexibility and adaptability to varying workloads deteriorate
Solution Approach 1:
The patent applies dynamics by transitioning from fixed capacity ceilings to dynamic capacity ceilings that automatically adjust based on real-time system conditions. The dynamic ceiling is calculated using a formula that incorporates current resource consumption, workload metrics, and historical data, allowing the system to adapt to varying workloads while maintaining resource management compliance. This resolves the contradiction by making the ceiling flexible rather than rigid.
Solution Approach 2:
The patent changes the parameter of capacity ceiling from a static value to a dynamic value that varies with system conditions. By introducing parameters such as current resource consumption, workload intensity, and time-based factors into the ceiling calculation, the system can maintain compliance while adapting to different operational states. This parameter transformation directly addresses the contradiction between fixed compliance and flexible adaptability.
2Power
If processor allocation is optimized for peak workload, then maximum processing capacity is improved, but resource utilization during low workload periods deteriorates
Solution Approach 1:
The patent implements dynamic processor allocation that adjusts capacity ceilings in real-time based on actual workload demands. During peak periods, the system allocates maximum processing capacity to meet high-demand requirements, while during low-utilization periods, it reduces allocation to optimize resource efficiency. This dynamic adjustment resolves the contradiction by preventing both over-provisioning and under-provisioning of processing resources.
3Device complexity
If static resource allocation is used to simplify management, then system complexity is reduced, but performance under varying workloads deteriorates
Solution Approach 1:
The patent implements a self-adjusting resource allocation system that automatically monitors system conditions and modifies capacity ceilings without manual intervention. The system uses embedded algorithms to detect workload patterns and autonomously optimize resource distribution, eliminating the need for complex manual management while maintaining high performance. This self-service approach resolves the contradiction by providing automated optimization without increasing operational complexity.
Data Source
AI summary
A method for monitoring the use capacity of a partitioned data-processing system, the system being configured to have a plurality of logical partitions sharing common physical resources, involves limiting access to the processor resources of partitions Pi in accordance with the value of the parameters DC_Pi(t) for setting an upper limit on the capacity of each of the partitions Pi, which are variable over time and are recalculated periodically. The values DC_Pi(t) are recalculated periodically in accordance with: the result NICRP(t) of measuring the instantaneous consumption level of the processor resources of all the partitions Pi; the result NICTi(t) of measuring the instantaneous workload level of each of the partitions Pi; the upper limit state of each of the partitions Pi; and at least one “Kuser” parameter determined by the user.