Parallel Rebuild Control for Storage Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The time required for the rebuild process in storage systems has increased with the growth in storage capacity, necessitating a solution to reduce this time while minimizing the impact on host I/O processes.
Innovation Solution
A storage system with a rebuild control unit that splits the rebuild process into partial processes, including data read and write operations, and executes these in parallel across multiple storage devices, allowing for efficient data reconstruction and distribution across spare and replacement storage devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the rebuild process is performed sequentially on storage devices, then the impact on host I/O processes is minimized, but the rebuild time increases significantly
Solution Approach 1:
The rebuild process is divided into multiple segments corresponding to different storage devices. Each segment is processed independently and in parallel, allowing the rebuild operation to proceed simultaneously across multiple devices without interfering with host I/O operations. This segmentation enables the system to achieve both fast rebuild times and maintained host performance.
2Quantity of substance
If storage capacity is increased to meet growing data needs, then more data can be stored, but the rebuild time increases proportionally
Solution Approach 1:
Large storage capacities are divided into multiple smaller segments across different storage devices. Each segment can be rebuilt in parallel, so the total rebuild time does not increase linearly with storage capacity. Instead, the parallel processing maintains relatively consistent rebuild times even as capacity scales up.
Solution Approach 2:
The rebuild process transitions from a single-dimensional sequential approach to a multi-dimensional parallel approach by utilizing multiple storage devices simultaneously. This dimensional change allows the system to handle larger capacities without proportionally increasing rebuild time.
3Loss of time
If multiple rebuild processes are executed simultaneously, then the total rebuild time is reduced, but the system complexity and resource management difficulty increase
Solution Approach 1:
The system implements self-service mechanisms where storage devices automatically coordinate their rebuild operations. Each device independently manages its own rebuild segment, reducing the need for complex centralized coordination and simplifying the management of parallel rebuild processes.
Data Source
AI summary
A rebuild control unit splits a rebuild process into partial processes and instructs an access processing unit to execute the partial processes. In the rebuild process, data identical to data recorded in a first storage device that constitutes RLU #1 is generated based on data read from, among storage devices constituting RLU #1, storage devices other than the first storage device and written to a spare storage device included in a storage system. Each partial process includes a combination of an operation of reading data from a split range created by splitting a data readout target range into fixed-size intervals and an operation of writing data in another storage device based on the data read from the split range. In response to the instruction from the rebuild control unit to execute the multiple partial processes, the access processing unit executes the instructed partial processes in parallel.