NVMe Technology in Data Centers
End users continue to rely on RAID for server-level primary storage protection. Crucial applications of RAID include cost-effective protection from drive failure for boot volumes, primary data storage and database applications.
Previously, NVM Express™ (NVMe™) device use has been confined to high-performance applications due to their high cost. However, as these costs have started to decrease, NVMe drives for traditional use models have emerged for use in enterprise data centers and are quickly becoming the preferred storage media for large capacities.
In addition, these other ecosystem challenges to using NVMe as primary storage have been removed:
- Enterprise reliability (write endurance)
- Serviceability (hot plug, surprise plug)
- Standardization (Universal Backplane Management)
- Technology for ubiquitous connectivity of SAS/SATA/NVMe (tri-mode)
The rising popularity of NVMe drives in the enterprise server space means that RAID for NVMe SSDs is quickly becoming a mandatory offering for enterprise servers.
Options for RAID in NVMe Ecosystems
Option 1 - Software RAID Architecture
As seen with this option, software RAID utilizes the in-box NVMe device driver on the host to access NVMe drives directly or via a switch where necessary. Utilizing the inbox driver provides the lowest possible latency path from the host to the drives and superior I/O performance when compared to traditional hardware RAID. A disadvantage of software RAID is the consumption of expensive compute and memory bandwidth on the host.
Option 2 - Single-Path RAID Architecture
For this option, single-path hardware RAID relieves the parity generation burden from the host. The use of single-path hardware RAID funnels all data to the RAID controller hardware prior to placing them on the submission queue of the drive. Directing all data through the RAID controller adds latency for read/write operations, even when intelligent bypass options are available to reduce unnecessary steps for various traffic patterns.
Option 3 - Multi-Path RAID Architecture
Multi-Path Drivers and the Path to Standardized Drivers
NVMe SSDs offer high IOP and low latency performance, which need to be preserved when using data protection services such as RAID. Traditional data protection has relied on either a software solution, which consumes valuable CPU resources, or a single-data path RAID controller with larger-than-optimal latency values. As the next generation of Flash media drives emerges, redesigning traditional data protection architectures is required.
All interactions with the multi-path driver present NVMe SSDs as SCSI devices. A standardized driver might interface directly to NVMe drives via the PCIe® switch when offloading is not required; or it might interface to a storage controller that presents itself as a logical NVMe drive that is attached to the switch.
The NVMe-based storage ecosystem, coupled with lowering costs, renews demand for expanded use of NVMe-based devices beyond limited, high-performance applications. While demand for offloaded hardware RAID acceleration and rich data services has not disappeared, traditional data path management needs to transition to new, optimized paths to maintain relevance in the NVMe market. Architectures based on our solutions deliver on this promise of optimized data paths for protected storage. Our SmartROC RAID-on-Chip controllers are versatile, fully featured and secure server storage solutions. Learn how to take advantage of their advanced embedded switch technology, innovative driver support and NVMe capabilities for creating the next generation of enterprise and cloud data centers.
