Storage I/O Routing via NVRAM Buffering for Write Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems face challenges in efficiently managing data across multiple storage devices, particularly in ensuring data persistence and reliability in the event of power failures or device failures.
Innovation Solution
The proposed solution involves a storage system architecture that utilizes a combination of non-volatile random access memory (NVRAM) devices and storage drives, with storage array controllers managing data across these devices. This architecture includes features like direct-mapped flash storage, zone-based storage management, and erasure coding for data redundancy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is written directly to storage drives, then data persistence is ensured, but write performance is reduced and latency increases
Solution Approach 1:
The patent introduces NVRAM as an intermediary buffer between the host system and storage drives. Control information is first written to NVRAM (providing fast write performance), then asynchronously transferred to storage drives (ensuring persistence). This mediator resolves the contradiction by decoupling the immediate write operation from the persistent storage operation.
Solution Approach 2:
The system performs preliminary writing of control information to NVRAM before final commitment to storage drives. The NVRAM serves as a preliminary storage location that provides immediate acknowledgment of write operations, while the actual persistent storage occurs subsequently, improving overall write performance without sacrificing data persistence.
2Reliability
If data is written directly to storage drives, then data reliability is improved, but access latency increases
Solution Approach 1:
NVRAM acts as a time-buffering intermediary that absorbs the time difference between fast host write operations and slower storage drive operations. Control information is quickly staged in NVRAM, allowing host systems to proceed without waiting for slow storage drive operations, thus reducing access latency while maintaining data reliability through eventual commitment to drives.
Solution Approach 2:
The system performs preliminary staging of control information in NVRAM before final storage drive operations. This preliminary action in fast NVRAM memory eliminates the time penalty from slow storage drive operations, as the host system receives immediate acknowledgment while the actual persistent storage occurs in the background.
3Productivity
If NVRAM is used as a buffer, then write performance is improved, but device complexity increases
Solution Approach 1:
The NVRAM device performs multiple functions: it serves as a write buffer for control information, a cache for frequently accessed data, and a persistence layer for critical metadata. By consolidating these functions into a single NVRAM component, the system achieves improved write performance without proportionally increasing complexity, as one device fulfills multiple roles.
Solution Approach 2:
The patent merges the buffer function, cache function, and persistence function into a single NVRAM component rather than using separate devices for each function. This consolidation improves write performance by providing a unified fast-write path while reducing the overall complexity that would result from multiple separate buffer, cache, and persistence devices.
4Stability of the object's composition
If control information is managed centrally, then data consistency is improved, but system scalability is reduced
Solution Approach 1:
The patent segments control information management by device type, with each storage device (NVRAM, storage drives) having dedicated control information specific to its operation. This segmentation allows each device to be managed independently with device-type-specific I/O paths, improving data consistency for each device while enabling the system to scale by adding different device types without requiring centralized reconfiguration.
Solution Approach 2:
The system dynamically selects I/O paths based on device type and operation characteristics. Rather than a fixed centralized control architecture, the system adapts its control information management approach according to the specific device being accessed, enabling both data consistency through device-appropriate protocols and scalability through flexible path selection.
Data Source
AI summary
Writing data in a storage system that includes a first type of storage device and a second type of storage device, including: selecting, for one or more unprocessed write requests, a target storage device type from the first type of storage device and the second type of storage device; issuing a first group of write requests to the first type of storage device, the first group of write requests addressed to one or more locations selected in dependence upon an expected address translation to be performed by the first type of storage device; and issuing a second group of write requests to the second type of storage device, the second group of write requests addressed to one or more locations selected in dependence upon a layout of memory in the second type of storage device.