Two-Tiered Cache Retention Management via Dynamic Track Policies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In two-tiered cache management systems, existing technologies face challenges in efficiently managing the retention time of tracks in Dynamic Random Access Memory (DRAM) and Storage Class Memory (SCM) caches, leading to potential cache overflow due to strict adherence to minimum retention times and varying access patterns, which can result in decreased cache hit ratios and increased processor intensity for garbage collection.
Innovation Solution
Implementing a method that receives minimum retention time indications from host applications for both DRAM and SCM caches, allowing for dynamic adjustment of retention times based on access patterns and cache utilization, preventing cache overflow while ensuring that tracks with specified retention times are maintained for the required duration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If strict minimum retention time is enforced for all tracks in cache, then track retention reliability is improved, but cache overflow occurs and cache hit ratio decreases
Solution Approach 1:
The patent applies different minimum retention time policies to different tracks based on their access patterns. Tracks with high access patterns receive extended retention times to maintain reliability, while tracks with low access patterns are subject to standard retention policies, allowing them to be evicted more aggressively to prevent cache overflow and maintain cache hit ratio.
Solution Approach 2:
The system dynamically adjusts retention time policies based on real-time access pattern analysis. The cache management apparatus monitors track access patterns and modifies retention time assignments accordingly, transitioning from static retention policies to dynamic, adaptive policies that balance reliability and cache efficiency.
2Reliability
If minimum retention time is extended to maintain track reliability, then track retention reliability is improved, but processor intensity for garbage collection increases
Solution Approach 1:
The patent differentiates retention time requirements by track access patterns, applying extended retention only to tracks that demonstrate high access patterns. This localized approach ensures reliability where needed while minimizing the overall garbage collection workload by not extending retention uniformly across all tracks.
Solution Approach 2:
The system applies minimum retention time extension partially only to tracks with high access patterns rather than universally to all tracks. This partial application reduces the excessive garbage collection intensity that would result from universal retention extension, while still maintaining reliability for critical tracks.
3Productivity
If cache retention policy is made more aggressive to prevent overflow, then cache management efficiency is improved, but track retention reliability decreases
Solution Approach 1:
The patent implements differentiated retention policies where high-access tracks receive protective retention extensions while low-access tracks are subject to aggressive eviction. This local quality approach allows aggressive cache management for most tracks while maintaining reliability for critical tracks with high access patterns.
Solution Approach 2:
The cache management apparatus continuously monitors track access patterns and uses this feedback to adjust retention time assignments. Tracks demonstrating high access patterns receive extended retention in response to feedback, while tracks with declining access patterns are subject to more aggressive eviction, balancing efficiency and reliability dynamically.
Data Source
AI summary
An indication is received from a host application of a first minimum retention time in a cache comprising a first type of memory and a second type of memory for a first plurality of tracks, wherein the first minimum retention time is not indicated for a second plurality of tracks. Based on the first minimum retention time, a second minimum retention time is set for the first plurality of tracks for the first type of memory and a third minimum retention time is set for the first plurality of tracks for the second type of memory. A track of the first plurality of tracks is demoted from the first type of memory, in response to determining that the track is a least recently used (LRU) track in a LRU list of tracks in the first type of memory and the track has been in the first type of memory for a time that exceeds the second minimum retention time.