Cache Policy Engine Using File-Level Heuristics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The performance bottleneck in computer systems due to the lagging speed of hard-disk drives compared to central processing units is exacerbated by the high cost of faster storage devices, necessitating more effective caching solutions.
Innovation Solution
Implementing a cache policy engine that utilizes file-level heuristics to cache and evict data in non-volatile memory, improving access rates by considering file operations and metadata, thereby enhancing system performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If LBA-level caching schemes are used to accelerate I/O data transfer, then access speed is improved, but cache hit rate is insufficient due to lack of file-level context
Solution Approach 1:
The patent transitions from LBA-level caching (single dimension) to file-level caching (adding file metadata dimension). The cache policy engine incorporates file-level heuristics such as file access patterns, file types, and metadata to make caching decisions, thereby improving cache hit rates while maintaining fast access speeds.
Solution Approach 2:
The cache policy engine continuously monitors file-level access patterns and metadata, using this feedback to dynamically adjust caching decisions. By analyzing file access behavior and updating cache policies based on observed patterns, the system optimizes both cache hit rates and access speeds adaptively.
2Speed
If non-volatile memory is used as cache, then access rate is improved, but system complexity increases due to cache management overhead
Solution Approach 1:
The cache policy engine autonomously manages caching decisions by analyzing file-level heuristics and access patterns. The system self-adjusts cache policies without requiring complex external control mechanisms, thereby improving access rates while keeping management complexity manageable through automated decision-making.
Solution Approach 2:
The patent changes the parameters used for cache management from basic LBA addresses to comprehensive file-level parameters including metadata, access patterns, and file types. This parameter enrichment enables more intelligent caching decisions that improve access rates while the automated policy engine keeps management complexity controlled.
3Productivity
If faster storage devices are adopted, then I/O bottleneck is reduced, but cost increases significantly
Solution Approach 1:
The patent introduces non-volatile memory as an intermediary caching layer between hard-disk drives and the CPU. This mediator provides fast access to frequently used data without requiring the entire storage system to be upgraded to expensive fast storage devices, thereby improving I/O throughput while controlling costs.
Solution Approach 2:
Instead of upgrading the entire storage system to expensive fast storage, the patent applies partial action by caching only the most frequently accessed data in non-volatile memory. This selective approach improves overall system productivity while avoiding the excessive cost of replacing all storage devices.
Data Source
AI summary
A method and system to perform caching based at least on one or more file-level heuristics. The caching of a storage medium in a caching device is performed by a cache policy engine. The cache policy engine receives file-level information of input/output access of data of the storage medium and caches or evicts the data of the storage medium in the caching device based on the received file-level information. By utilizing information about the files and file operations associated with the disk sectors or logical block addresses of the storage medium, the cache policy engine can make a better decision on the data selection of the storage medium to be cached in or evicted from the caching device in one embodiment of the invention. Higher cache hit rates can be achieved and the performance of the system utilizing the cache policy engine is improved.