Ad Image

SSD Predictions: What’s Next for Enterprise SSD in 2023 & Beyond

SSD Predictions for 2023

SSD Predictions for 2023

This is part of Solutions Review’s Premium Content Series, a collection of contributed columns written by industry experts in maturing software categories. In this submission, Phison Product Manager Louis Liao offers SSD predictions for 2023 and beyond for enterprise organizations.

SR PremiumAbout 15 years ago, enterprise solid-state drives (SSDs) began to be considered a separate category of drive compared to commercially available SSDs used by consumers. Today, the drives are becoming increasingly common in data centers and other enterprise applications across the world. In this article, we explore trends and industry advancements that will become more prevalent through the next several years.

Security will continue to be a primary focus for enterprise data centers and infrastructure, including SSDs.

Cyber-attacks are consistently on the rise year over year, and 2023 will be no exception. Data security will remain a key priority for enterprises and their data centers. KPMG’s 2022 Fraud Outlook report found that 84 percent of survey respondents say their cyber risk will grow over the next year.

SSDs are typically defended well enough by layered security in the operating system. However, it is possible that attackers could get to the data stored on the drive if they are able to break through defense-in-depth countermeasures. To strengthen the protection of their SSDs, the most security-conscious enterprise customers will increasingly look for drives that offer a variety of security features, from self-encryption to secure erase and sanitize capabilities.

Another security capability on the rise is cryptographic device identity and attestation. Think of this as a cryptographically protected health check. It allows the SSD to perform an integrity verification on the firmware, hardware, and even important data on the drive. If anything changes, it is automatically detected and reported.

Some drives even have their own built-in defense countermeasures that can shut down automatically when they detect that they’ve been moved or disconnected and when they identify suspicious access patterns. These self-defending SSDs have a CPU dedicated to security that runs its detection and defense programs. In addition to shutting down, the SSDs’ defensive measures include alerting a support app, locking the drive with cryptographic keys, and erasing the drive completely.

Download Link to Data Storage Buyer's Guide

SSD Predictions for 2023

SSD Advance to Support High-Performance

The rising popularity of advanced technologies such as Artificial Intelligence (AI) and the Internet of Things (IoT) will drive increased demand for data storage (and therefore SSDs) that can deliver the high performance and low latency that support and enable complex high-capacity workloads. Enterprises will need SSDs that are ever faster, smarter, and denser for high-performance computing of all kinds. Both 32 TB and 64 TB SSDs are available now. Expect to see petabyte-scale SSDs starting at 128 TB in the near future.

SSD manufacturers are exploring potential technological breakthroughs to meet those performance demands. Some of the newer technologies that will increasingly support SSDs and boost their reliability and performance, include:

  • Compute Express Link (CXL): This technology piggybacks on the fast PCIe interconnect and provides cache coherency between host servers and other devices, so memory across the entire system is treated as a single accessible pool. It enables the system to access SSD data through the CPU’s existing memory hierarchy, which can shave off up to 20 microseconds of latency from bypassing the file system and its accompanying layers.
  • NVMe over Fabric (NVME-oF): This protocol can facilitate data transfer between a host server and an SSD or other device, which is more and more helpful as data volumes get larger and aren’t within 100 feet of physical distance of each other. NVMe-oF enables enterprises to get high-performance storage with low latencies, similar to direct attached storage.
  • Zoned Namespace (ZNS): A fairly new command set in NVMe, ZNS manages zones on an SSD so data is written to it in sequence. That eliminates the need for the drive to clean up old data later, and can significantly reduce latency, improve throughput and prolong SSD life.

Other technological advancements growing in popularity include:

  1. AI-powered NAND feature management, which optimizes NAND parameters for optimal read voltage.
  2. Wider channels between the SSD controller and NAND to boost IOPS for random reads.
  3. I/O path connections between the SSD and storage system enhance overall system performance.
  4. Key-value storage, which replaces the traditional block interface with a native database interface. This allows databases to be easily partitioned and enables horizontal scalability at a level other databases can’t match.

SSDs and Data Storage Begin to Take Full Advantage of Gen5 Performance

Gen5 products have begun entering the market, but the technology hasn’t been used to its fullest potential. In the coming year, as products such as AMD Genoa and Intel Sapphire Rapids go into mass production, experts estimate we’ll see enterprise SSDs get further upgrades to support PCIe Gen5 transfer rates up to 32 gigatransfers per second. These high transfer rates will eventually become the standard as enterprises deploy AI, ML, big data, and IoT applications that require high-speed computing capabilities.

SSD capacity will also need to increase to support those demanding applications. Industry analysts expect that due to the oversupply (and therefore price drop) of NAND flash, more enterprises will opt for SSDs with capacities of 4 TB or more. In fact, due to the intrinsic reliability of NAND storage and fast RAID rebuild times, data centers are becoming more comfortable with ultra-high-density drives.

Higher-Performance Technology Paves the Way for EDSFF SSD Form Factors

As Gen5 becomes commonplace, it will have an effect on SSD form factors. Experts estimate that the increased performance of PCIe Gen5 will result in an increase of EDSFF SSDs that could potentially replace up to half of the now-standard 2.5-inch SSDs used in enterprise storage and servers.

The EDSFF is a family of form factors that were designed specifically to meet the needs of data center storage. These SSDs offer enhanced performance, manageability, scalability, thermal and power management, and capacity than other form factors.

In many cases, capacity is the key to switching up form factors. As Gen5, AI, ML and high-performance computing—all of which generate and/or run on massive amounts of data—become more prevalent, the need for high-capacity drives will increase accordingly.

Consider the E1.L form factor, for instance, or the “ruler” as it’s often called because of its length (unsurprisingly, the L stands for long). It’s becoming more popular now to support hyperscalers and high-capacity use cases. Big players in the industry are interested in E1.L for three reasons:

  • Enterprises have become comfortable with large drive densities, and SSDs have taken over most drive slots previously held by hard disk drives (HDDs). While HDD vendors have been pushing high-capacity 16 TB to 20 TB HDDs, it’s really the only card they have left to play—and they still can’t beat the high performance and low latency of SSDs.
  • A CPU only has so many PCIe lanes, so the idea of having lots of small drives of 2 TB or 4 TB is becoming obsolete. Instead, the chassis will dedicate 50 to 80 PCIe lanes to storage and those drives need to be big to support today’s complex workloads.
  • SSDs are substantially more reliable than HDDs and they have very fast read/write bandwidth. Data centers are no longer worried about long rebuild time or having SSDs die in clusters.

The diagram below is to scale and shows the differences between the new EDSFF form factors and the legacy 3.5” and 2.5”. (Currently, there is no market interest in using the 3.5” form factor for anything other than HDD).

Source: Phison

The U.2 form factor is being replaced by E3.S because the U.2 connector is iffy at Gen5 and won’t work for Gen6. The EDSFF connector is designed for Gen5 and beyond.

The market is currently split between the different EDSFF form factors, which is a good thing, as the point of EDSFF in the first place was to provide different options to different users. The three most commonly used EDSFF are:

  1. E1.S: An improved M.2, this form factor is used primarily by second-tier storage appliances focused on lower density per SSD.
  2. E1.L: Can support the highest density while still allowing for simple front access.
  3. E3.S: Designed as a replacement for U.2, this form factor is used primarily by tier-1 OEMs with customers that want large density. Though the board is slightly bigger than the U.2, it will likely provide the same densities.

Sustainability Become Commonplace

Today, data center emissions represent 2 percent of the total global greenhouse gas emissions. Across the world, enterprises are beginning to focus more and more on sustainability initiatives that will help them reduce their carbon footprint and save energy. A report by Gartner found that enterprise spending on sustainability has increased by an average of 5.8 percent since 2017.

Increasing sustainability requires enterprises to make their processes more efficient, and also to implement more energy-saving hardware. Gartner vice president and analyst Frances Karamouzis said, “Targeting areas such as data-center power consumption is low-hanging fruit because it’s easy to measure.”

Enterprises will increasingly look for SSDs that consume less power than others, such as the Phison X1 SSD that still features best-in-class performance but uses less power than competitor’s products. The

less energy an SSD uses, the less heat it generates—which in turn reduces the need for cooling. That can help enterprises save significantly on energy costs.

Some enterprises will even turn to high-density, all-flash data centers to realize great energy savings and reduce their carbon emissions. Currently, cost is the primary reason enterprises are hesitant to adopt all-flash systems. However, the SSD cost per GB is expected to drop by half with the arrival of the 1 terabit (Tb) TLC NAND die in 2023 and drop again by half with a 2 Tb NAND die after 2026.

Enterprises Will Move Data to the Edge

As the use of AI, smart IoT devices and sensors, cloud applications, and data analytics become more commonplace; enterprises are increasingly turning to edge computing to leverage data faster and more efficiently closer to where it’s collected or generated. A report by Gartner found that several years ago, less than 10 percent of enterprise-generated data was created and processed outside a traditional centralized data center or cloud—but that figure could reach 75 percent by 2025.

Edge servers and edge data centers require SSDs that can deliver high performance and speed with low latency. And because “the edge” could be a remote cell tower in the middle of the desert, an outdoor seismometer at an arctic research outpost, or even a sensor on the International Space Station, SSDs used in edge systems need to be physically hardy—even ruggedized—as well as ultra-secure and able to withstand extreme weather or vibrations.

SSDs used at the edge don’t necessarily need to have massive storage capacity; more important is the ability to process and analyze time-sensitive information before sending results and other data on to a centralized system. Because that requires speed and agility, some enterprises are more interested than ever in NVMe-oF technology and storage-class memory.

Download Link to Data Storage Buyer's Guide

Louis Liao
Latest posts by Louis Liao (see all)

Share This

Related Posts