Tracing the Evolution of Data Center SSD Form Factors

An interview with Wayne Allen, Senior Director of Engineering, Data Center, at Solidigm

With more than 20 years of experience in the storage and technology industries, Wayne has overseen the development, standardization, and adoption of numerous technologies found in today's data centers. Below are his insights from a boots-on-the-ground career focused around the build and evolution of SSDs.

Since its widespread adoption in the early 2000s, SSD storage technology has—quite literally—changed shape. Advancements in engineering, together with increasing demands from users on both the client and data center side, have driven a series of innovations to improve everything from speed to capacity. To accommodate these innovations, SSD form factors have evolved, too.

SSDs were developed as an alternative to an older type of drive: HDDs (Hard Disk Drives). Where HDDs use mechanical spinning disks and a read/write head to access data, SSDs simply use memory chips—making them faster, quieter, more energy-efficient, and more durable. However, in the early days of widespread adoption, SSDs needed to connect to the PC or server motherboards in the same way that their predecessors did: through an interface called SATA.

It wasn’t just the interface that early SSDs mimicked. To fit into motherboard configurations at the time, these early SATA SSDs also adopted the shape of their HDD competitors. Thus, the 2.5-inch SATA SSD was born, offering a drop-in alternative to competing HDDs.

Before diving into the evolution of SSD form factors for the data center specifically, we’ll take a quick detour to the client side, where the 2.5-inch SATA form factor has streamlined significantly over time. Note that some of these form factors are used in data center and enterprise applications, though their most common adoption is in client devices such as laptops, tablets, and other personal devices.

One such popular form factor for compact devices is mSATA, which emerged in 2009 as a much smaller alternative to the 2.5-inch SATA form factor that still utilized the same interface. Approximately one-third the size of their 2.5-inch predecessors, mSATA SSDs quickly became a competitive alternative as client devices themselves—namely, laptops—began to shrink.

Following mSATA, a new drive was introduced with technology that greatly improved nearly every aspect of storage performance. Named M.2, it featured an ultra-slim footprint with an array of lengths to integrate into different systems. It also showcased a new interface—PCIe (short for PCI Express)—that offered significantly faster data transfer speeds than the SATA interface. High-performing, compact, and incredibly versatile, M.2 has become the preferred form factor for nearly all space-efficient client devices.

On the data center side, the evolution of data center SSD form factors directly parallels the performance demands of the industry—which are distinct from those on the client side.

The recent form factor history of data center SSDs also starts with the SATA and SAS interface. And like client SSDs, this history takes shape based on that same heritage—even as interfaces have evolved from SATA/SAS to PCIe.

Wayne explains, “When the industry started to go towards PCI electrical connections to the server, they basically took a 2.5-inch form factor.” Like all SSDs in the early 2000s, data center SATA SSDs adopted the easily implementable 2.5-inch form factor—and as interface technology progressed, that form factor stuck around.

Enter U.2. Designed as a drop-in replacement for 2.5-inch SATA SSDs, U.2 SSDs feature a newer PCIe interface with a similar form factor as their SATA predecessors, but slightly bulkier. They’re also much bulkier than mSATA and M.2 drives—though this bulkiness affords them more impressive spec sheets in key performance metrics. U.2 has become one of the preeminent form factors for data center applications, where performance is prioritized, and space is more readily available.

Though space is more abundant in a server than in a slim laptop, it’s not unlimited. As data storage demands increased over time, data storage engineers and their enterprise clientele began searching for ways to improve space efficiency in servers designing a form factor specifically around NAND flash chips. Wayne, as part of his two-decade career at Intel, was at the forefront of this search, developing new form factors by thinking both inside and outside of the server box.

“We came up with all sorts of hair-brained, wild ideas—you know, from like soldering down the SSD controller to the motherboard and just having DIMMs of NAND, kind of like DRAM. We ended up coming up with this form factor dubbed ‘The Ruler’ as it was roughly 12 inches long and a little more than an inch tall.”

On the performance side, the ruler truly measured up—saving space versus its U.2 cousin, enabling enterprise users to install more drives per server, and increasing storage. As the new form factor was shopped around, it led to the formation of an industry-wide standards consortium called the Enterprise and Data Center Standard Form Factor (EDSFF). With what Wayne describes as “slight adjustments,” the ruler was standardized and led to a family of EDSFF variants based on the same connector.

As innovations became standardized with EDSFF, two new form factors led the way in a relatively short timeframe. The ruler became E1.L — (L standing for long) followed by the E1.S (S standing for short), as this form factor doesn’t reach as far back into a server. The E1 is a form factor that fits into a 1U server. In the simplistic picture below, alternative form factors reduce the number of drives that can fit in the system. In some 1U systems that have constrained front panel space due to front panel I/O and other factors, E1 enables a greater number of drives to fit in the system compared to alternative form factors.

Most recently unveiled under the EDSFF banner are the new E3 form factors: E3.S, E3.S 2T, E3.L, and E3.L 2T. As the primary replacement for the U.2 form factor (likely in the PCIe Gen6 timeframe), these EDSFF variants have more versatility and increased capacities, better thermal efficiency, and exceptional read/write speeds.

The newer E3 form factors offer improved total cost of operation (TCO) right out of the gate in many use cases—as comparable or superior storage can be achieved with fewer drives than competing U.2 configurations. The E3 drives present a significant, future-proofed upside for data centers with the electrical characteristic needs of newer/faster PCIe generations.

Learn more about how Solidigm’s drives can help you get the right performance at a substantially lower TCO for mainstream and read-intensive workloads in content delivery networks, AI, data pipelines, object storage, and more at www.solidigm.com.