Memory Manager Voting Algorithm for Data Eviction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In scenarios where physical memory cannot hold all data, existing technologies lack an efficient method to select eviction candidates based on multiple criteria, leading to suboptimal resource allocation and management.
Innovation Solution
A resource manager, specifically a memory manager, uses a voting algorithm to determine eviction rankings based on multiple criteria, allowing for the selection and reassignment of memory blocks by aggregating favorability measures across multiple rankings, and adapts criteria dynamically or statically to optimize resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical memory is used to store all accessed data, then data availability is improved, but memory capacity is exceeded and resource allocation becomes suboptimal
Solution Approach 1:
The patent extracts the decision-making process for eviction from simple LRU logic and removes least valuable data from memory based on multiple criteria evaluations. This allows the system to maintain only essential data in physical memory while extracting non-essential data for eviction, thereby resolving the contradiction between data availability and memory capacity constraints.
Solution Approach 2:
The patent introduces an intermediary decision-making mechanism that evaluates multiple eviction criteria (access patterns, data value, resource requirements) before selecting eviction candidates. This intermediary layer between data storage and memory management enables optimal resource allocation by mediating which data should be retained or evicted, thus balancing data availability with limited memory capacity.
2Productivity
If multiple eviction criteria are used to select eviction candidates, then resource allocation optimality is improved, but decision-making complexity increases
Solution Approach 1:
The patent segments the eviction decision-making process into distinct evaluation criteria (access patterns, data value, resource requirements) that can be independently assessed. Each criterion contributes separately to the overall eviction decision, allowing the system to achieve optimal resource allocation through modular evaluation rather than monolithic complex processing.
Solution Approach 2:
The patent changes the parameters of the eviction decision from simple LRU time-based metrics to multiple configurable criteria with adjustable weights. This parameter transformation enables the system to adapt to different workloads and optimization goals while maintaining a structured decision-making framework that manages complexity through standardized parameter evaluation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a memory manager (12). The memory manager determine a plurality of eviction rankings (18,20). Each of the plurality of eviction rankings assigns a position within the eviction ranking to each of a plurality of eviction candidates based on at least one eviction criterion (24,28). The memory manager selects one of said eviction candidates by applying a voting algorithm to the plurality of eviction rankings and causes the selected eviction candidate to be evicted.