Disaggregated Storage Host NVRAM Buffer Redundant Writes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional storage systems face inefficiencies in data management and storage operations, particularly in handling flash drives without integrated storage controllers, leading to redundant write operations and reduced reliability.
Innovation Solution
A direct-mapped flash storage system where the operating system initiates and controls processes, offloading device management from flash drives by using non-volatile random access memory (NVRAM) as a buffer for write operations, and employing erasure coding and mirroring for data redundancy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If flash drives without integrated storage controllers are used, then device complexity is reduced, but storage system reliability deteriorates due to redundant write operations
Solution Approach 1:
The patent introduces a host system with NVRAM as an intermediary between the host controller and flash drives without controllers. The NVRAM buffer stores write data temporarily, allowing the system to coordinate write operations across multiple flash drives and prevent redundant writes, thereby maintaining reliability while using simpler flash drive hardware.
Solution Approach 2:
The storage controller functionality is extracted from the flash drives and centralized in the host system. This allows the host to manage write operations globally across all flash drives, preventing redundant writes by coordinating which drives receive write commands, thus improving reliability while keeping individual flash drives simple.
2Ease of operation
If traditional storage systems with integrated controllers are used, then storage management is simplified, but productivity decreases due to redundant write operations
Solution Approach 1:
The controller functionality is extracted from individual flash drives and consolidated in the host system. This centralization eliminates redundant write operations by allowing the host to track which drives have been written to, improving productivity. The host maintains a global view of storage state, simplifying coordination while removing wasted write operations.
Solution Approach 2:
The host system's NVRAM and controller logic serve multiple flash drives simultaneously, providing universal management across the storage array. This multi-functional approach allows a single host controller to manage many flash drives efficiently, avoiding redundant operations and improving overall storage productivity.
3Reliability
If redundant write operations are performed, then data redundancy is improved, but loss of energy increases
Solution Approach 1:
The host system implements feedback mechanisms by tracking which flash drives have received write commands and storing this state information in NVRAM. Before issuing write commands, the host queries this state to determine which drives still need data, preventing redundant writes and reducing the energy consumed by unnecessary write operations while maintaining data redundancy.
Solution Approach 2:
The system performs preliminary actions by maintaining a state record in NVRAM of which drives have been written to. This preliminary information allows the host to make informed decisions about subsequent write operations, avoiding redundant writes and reducing energy consumption before write operations are actually initiated.
Data Source
AI summary
Disaggregated storage systems for hyperscale deployments including retrieving workload metrics and a workload policy associated with a workload executing within a fleet of storage systems; accessing a fleet configuration of the fleet of storage systems; and generating, based on the workload metrics, the workload policy, and the fleet configuration, a fleet reconfiguration recommendation indicating a change to hardware installed on a storage system in the fleet of storage systems.