File System Auto-Tiering Across Heterogeneous LUNs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional tiered data storage systems are ineffective in managing file systems across multiple logical units of storage (LUNs) due to inefficient tiering policies, leading to underutilization of magnetic disk drive memory and exhaustion of flash memory capacity, as they rely solely on slice activity levels for data relocation.
Innovation Solution
Implementing a method that provisions LUNs from a heterogeneous storage pool comprising SSD and HDD tiers, applying distinct tiering policies to each LUN to move active files to high-performance SSDs and inactive files to cost-effective HDDs, using stub files and content files to maintain file system integrity while optimizing storage usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a tiering policy relocates slices based on slice activity level, then active slices are placed in flash memory for high performance, but magnetic disk drive memory becomes underutilized and flash memory capacity is exhausted
Solution Approach 1:
The patent segments the storage management at two levels: slice level (conventional approach) and file level (new approach). By dividing files into multiple slices and tracking file-level activity, the system can make more granular decisions about data placement, preventing both flash memory exhaustion and magnetic disk underutilization.
Solution Approach 2:
The patent applies different quality characteristics to different parts of the storage system by implementing file-level activity tracking and policy differentiation. Different files within the same LUN can have different tiering policies applied based on their specific activity patterns, allowing optimal placement of each file according to its actual usage characteristics rather than blanket slice-level policies.
2Productivity
If all slices are relocated to flash memory due to high activity from a few files, then active file access performance is improved, but storage cost increases and magnetic disk drive memory is underutilized
Solution Approach 1:
The patent changes the parameter of activity tracking from slice-level to file-level, and introduces policy differentiation as another variable. This allows the system to adjust data placement decisions based on actual file activity patterns rather than aggregate slice activity, optimizing the balance between performance and resource utilization.
3Ease of operation
If a single tiering policy is applied to all LUNs, then management is simplified, but different storage performance requirements of different LUNs cannot be met
Solution Approach 1:
The patent introduces dynamic policy assignment where LUNs can be assigned different tiering policies based on their specific requirements. The system can dynamically adjust which policy applies to which LUN, allowing flexible adaptation to different storage performance requirements while maintaining manageable complexity through automated policy selection or configuration.
Data Source
AI summary
A technique manages file data of a file system. The technique involves provisioning a first LUN with slice locations from a heterogeneous storage pool created from a solid state device (SSD) storage tier formed from SSD memory and a hard disk drive (HDD) storage tier formed from HDD memory. The technique further involves provisioning a second LUN with slice locations from the heterogeneous storage pool. The technique further involves, while different LUN level policies are applied to the first LUN and the second LUN, moving file data of a file system from the slice locations of the first LUN to the slice locations of the second LUN. Such a technique enables effective auto-tiering at the file level where active file data moves to a higher performance storage tier and inactive file data moves to a cost effective lower performance storage tier thus optimizing operation.