Application-Level Data Importance for Tiered Storage Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current storage systems lack the ability to distinguish between different types of data for optimal storage, leading to important data being frequently cycled out of the cache and inefficient use of high-performance media due to the lack of selective storage based on importance levels.
Innovation Solution
A system that includes an application server and a storage control module to assign importance levels to data, storing critical data in a high-performance storage medium (such as SSD) and less critical data in a lower-performance medium (such as HDD), with the option to use RAID storage arrays for redundancy and security, and a DRAM cache for improved access times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If all data is cached in DRAM to improve access performance, then data access speed is improved, but storage cost increases and cache capacity is limited
Solution Approach 1:
The patent segments data into different importance levels (critical, important, standard, low-priority) and segments storage media into different performance tiers (DRAM cache, SSD, HDD). This allows critical data to be placed in high-performance DRAM cache while less critical data resides in lower-performance but higher-capacity storage, resolving the contradiction between access speed and cache capacity.
Solution Approach 2:
The patent applies local quality by assigning different storage qualities to different portions of data based on their importance levels. Critical data receives high-quality storage in DRAM cache for fast access, while non-critical data is stored in lower-quality media, optimizing the overall system performance without requiring all data to be in high-performance media.
2Speed
If DRAM cache is used to store frequently accessed data, then data access performance is improved, but important data is frequently cycled out due to lack of distinction between data types
Solution Approach 1:
The patent applies preliminary action by pre-classifying data into importance levels before caching. This allows the system to proactively identify and retain critical data in DRAM cache, avoiding the need to cycle it out. The classification happens in advance, ensuring that important data is always available in high-performance storage when needed.
Solution Approach 2:
The system uses feedback from application-level prior knowledge about data importance to dynamically manage cache contents. By monitoring which data is marked as critical by applications, the system can adjust cache retention policies to keep important data in DRAM cache longer, improving both performance and reliability.
3Speed
If high-performance storage media is used for all data, then data access latency is reduced, but financial cost per unit of stored data increases
Solution Approach 1:
The patent segments the storage system into multiple tiers with different performance characteristics and costs (DRAM cache, SSD, HDD). By segmenting both data and storage media, the system can place only critical data in expensive high-performance media while storing less critical data in cheaper lower-performance media, reducing overall storage costs while maintaining necessary performance.
Solution Approach 2:
The patent changes the performance parameters of different storage media to match the importance levels of data. Critical data is assigned to high-performance media with low latency, while non-critical data is assigned to lower-performance media. This parameter matching optimizes the cost-performance ratio of the storage system.
Data Source
AI summary
A method for selectively storing data identified by a software application in higher performance media may include executing control programming for an operating system and a software application hosted by the operating system. The software application assigns a first importance level to a first portion of data and a second importance level to a second portion of data. A first portion of data having the first importance level assigned by the software application is stored in a first storage medium at the instruction of the operating system. A second portion of data having the second importance level assigned by the software application is stored in a second storage medium at the instruction of the operating system. The second storage medium has at least one performance, reliability, or security characteristic different from the first storage medium.