Understanding SSD Technology: NVMe, SATA, M.2

SSD Storage: SATA vs PCIe

Data buses transfer data within a system and when NAND-based SSDs first came out, it was clear to the industry that a new bus and protocol were necessary.

• The limitations of HDDs led to the adoption of Solid-State Drives (SSDs). SSDs were able to utilize the same infrastructure used for HDDs and brought much better performance via sequential and random performance. Although the SATA III specification supported speeds of up to 6 Gbps, hard drives were unable to fully utilize this bandwidth—whereas the new, commercially-available SSDs quickly began to saturate the interface.
• PCIe 4.0 doubled that to 32GB/s (256 Gbps) for x16, significantly increasing bandwidth (16 Gbps for x1). PCIe 4.0’s 16 lanes can transfer data up to 32,000MB/s. PCIe 5.0 again doubled the bandwidth, reaching 64GB/s (512 Gbps) for x16 (32 Gbps for x1), with a theoretical maximum transfer rate of 64,000MB/s. For comparison, the older SATA III standard tops out at only 600MB/s, making PCIe 5.0 over 100 times faster than SATA III, highlighting the massive performance gains offered by each new generation of PCIe.

The decision to leverage an existing higher-bandwidth bus technology replaced SATA protocols with PCIe technology. PCIe storage came before NVMe by a few years but since previous solutions were bottlenecked by older data transfer protocols such as SATA and AHCI, it couldn’t be used to its full potential until recent years. NVMe was the solution for the bottlenecks and removes limitations by offering low-latency commands and 64K queues. The multiple queues allow faster data transfers because data is written to SSDs in a scattered fashion using chips and blocks rather than being written on spinning disks like hard drives.