Dynamic Storage Allocation Between Log Structured File System and RAID
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In storage systems with log-structured writes, RAID modules need to reserve spare capacity for drive failures and rebuilds, but this can conflict with the log structured file system's requirement to maintain low utilization to minimize garbage collection costs and maximize write performance.
Innovation Solution
A method to identify and allocate storage space reserved by the file system for RAID rebuilds, allowing the RAID to utilize this space while ensuring the file system maintains low utilization, thereby optimizing performance and availability of spare capacity for rebuilds by performing garbage collection when needed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If RAID reserves spare capacity for drive failures and rebuilds, then reliability is improved, but the log structured file system cannot overprovision the log and maintain low utilization, worsening write performance
Solution Approach 1:
The patent implements dynamic capacity allocation where the log structured file system can temporarily borrow spare capacity from RAID when needed for log operations, and return it when RAID needs it for rebuilds. This dynamic sharing resolves the contradiction by allowing both systems to access the capacity as needed rather than having static reservations that force one system to sacrifice performance or reliability.
Solution Approach 2:
The system changes the parameter of capacity reservation from fixed to variable. The log structured file system can adjust its effective log capacity by borrowing from RAID's spare pool, allowing it to maintain low utilization and minimize garbage collection costs while RAID maintains its rebuild capability through dynamic replenishment of spare capacity.
2Loss of energy
If log structured file system overprovisions the log to keep utilization low, then garbage collection cost is reduced, but RAID loses spare capacity needed for rebuild operations
Solution Approach 1:
The patent merges the capacity pools of RAID and log structured file system into a shared resource. The log system can overprovision by drawing from the combined pool, reducing garbage collection costs, while RAID maintains its rebuild capability through dynamic allocation. This merging allows both systems to benefit from increased available capacity without permanently sacrificing either garbage collection efficiency or rebuild reliability.
3Reliability
If RAID maintains dedicated spare capacity for immediate rebuilds, then reliability is improved, but total storage capacity available for data storage is reduced
Solution Approach 1:
The system transitions from static dedicated spare capacity to dynamic shared capacity. RAID can maintain its rebuild capability by temporarily allocating capacity from the log structured file system when rebuilds are needed, rather than permanently reserving capacity that sits idle when not needed for rebuilds. This dynamic approach preserves reliability while maximizing total usable storage capacity.
Data Source
AI summary
A method, computer program product, and computer system for identifying, by a computing device, a first amount of storage space reserved by RAID for a RAID rebuild of a failed drive. A second amount of the storage space reserved by a file system may be identified, wherein the storage space may be shared between the RAID and the file system. The RAID rebuild of the failed drive may be performed. The first amount of the storage space may be allocated to the RAID from the second amount of the storage space reserved by the file system.