Open Channel SSD Pooling via FTL Daemon
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data center systems using traditional NAND SSDs face inefficiencies due to the inability of host devices with HDD interfaces to fully utilize SSD performance, leading to unpredictable data storage and increased wear on SSDs, especially when managing large-scale deployments of open-channel SSDs, which require significant system resources and overhead.
Innovation Solution
Implementing a method to manage a plurality of open-channel SSDs over a network by creating abstracted memory structures that allocate free pages across multiple SSDs, maintaining physical addresses, and assigning these structures to host devices, while also managing reliability and wear through mechanisms like bad block management, thereby reducing system overhead and improving efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional NAND SSDs with FTL are used, then host device integration is easy and FTL overhead is offloaded, but SSD performance cannot be fully utilized and data storage becomes unpredictable
Solution Approach 1:
The patent introduces an FTL daemon as an intermediary software component that runs on the host device. This daemon translates high-level host commands into low-level SSD commands, acting as a mediator between the host file system and the SSD controller. It maintains translation layers, manages block mappings, and coordinates with the SSD's native FTL to enable both easy integration and full performance utilization.
Solution Approach 2:
The patent segments the FTL functionality into two parts: a software FTL daemon running on the host device and the hardware FTL in the SSD controller. This segmentation allows the host to participate in FTL operations while the SSD maintains its native capabilities, enabling both compatibility and performance optimization simultaneously.
2Productivity
If open-channel SSDs without firmware FTL are used, then SSD performance can be fully utilized, but system overhead increases significantly as host OS must manage each SSD individually
Solution Approach 1:
The patent merges multiple open-channel SSDs into a single logical storage pool managed by a unified FTL daemon. Instead of the host OS managing each SSD separately (which creates significant overhead), the daemon consolidates management functions across all SSDs in the pool, reducing complexity while maintaining full performance utilization.
Solution Approach 2:
The FTL daemon serves multiple functions simultaneously: it acts as a translation layer for host commands, manages the storage pool across multiple SSDs, performs garbage collection, handles wear leveling, and coordinates with SSD controllers. This multi-functionality reduces the need for separate management systems for each SSD.
3Quantity of substance
If large-scale deployments of open-channel SSDs are implemented, then storage capacity increases, but system resources and overhead consume increasingly larger proportions
Solution Approach 1:
The patent merges many individual SSDs into a unified storage pool managed by a single FTL daemon instance. This consolidation allows the system to scale storage capacity without proportionally increasing management overhead, as the daemon handles multiple SSDs efficiently through pooled resource management.
Solution Approach 2:
The patent implements a virtualization layer that creates abstracted views of physical storage resources. The FTL daemon manages physical SSDs while presenting virtual storage interfaces to the host, allowing multiple host systems to share access to the same physical pool without each requiring dedicated management resources.
Data Source
AI summary
In one embodiment a plurality of open channel solid state drives (SSDs) are implemented over a network comprised of a network switch having a plurality of nodes, a remote host connected to a first node of the network switch, a metadata server connected to a second node of the network switch, and an abstracted memory structure comprised of at least part of one of the plurality of open channel SSDs. In one embodiment, the remote host is configured to communicate with the metadata server by issuing a command identifying data related to the abstracted memory structure. In another embodiment, the metadata server is configured to communicate with the remote host by responding to the command and identifying a physical address corresponding to the identified data.