New Urgency for More Disks-Per-Drive Needed in Nearline HDDs?

This is an Executive Brief by Trendfocus, Inc.:

More Disks-Per-Drive Needed in Nearline Has New Urgency?

With all HDD vendors on 9-disk platforms for current-gen nearline HDD products that support 16TB and 18TB CMR capacities, a question looms soon whether more disks-per-drive may be required to reach even higher capacities.

Both Western Digital and Seagate have commenced shipping low volumes of 18TB, 9-platter HDDs, supporting 2TB/disk; however, the means to reach these per-disk capacities have required employing various design choices and technologies.

Western Digital has turned to what it now calls EAMR, or energy-assisted magnetic recording – a white paper it published details the technology stop-gap on its development path to MAMR (microwave-assisted magnetic recording) and HAMR (heat-assisted magnetic recording) which can be downloaded.

EAMR, as described by Western Digital, applies energy to the write head via a head bias current to stabilize the write signal and to provide additional write field in order to magnetize, or write, harder to switch magnetic media which, in turn, enables higher density recording.

This is in contrast to MAMR and HAMR which apply energy to the magnetic media on the disk itself, temporarily lowering the energy required to change magnetization direction (or to write the disk). Energy application to the disk media is the long-term holy grail of HDD recording technologies as it enables the use of magnetic materials supporting ultra-high-density recording – magnetic material that could not be written at room temperature with conventional write heads, with or without the applied bias current approach.

Seagate has publicly discussed its plan to ship a 9-disk, 20TB HDD by the end of 2020 using HAMR.

However, CEO Dave Mosley during the company’s 4FQ20 earnings call in July stated: “We plan to offer 20TB HAMR drives to customers on a limited basis and it is part of our system solution to collect production and field data.”

While this would be a milestone achievement on the part of Seagate, all indications point to this product as not a mainstream, high-volume product to address the expected 20TB market that Trendfocus has forecasted to emerge later in 2021. The product nevertheless will enable Seagate to gain valuable experience in maturing HAMR for future high-capacity models.

MAMR and HAMR (what Trendfocus is calling full energy-assisted recording, as opposed to Western Digital’s current EAMR approach) are difficult technologies to commercialize and despite years of development and investment, both continue to suffer delays.

Trendfocus’ 2CQ20 revised long-term forecasts, published in August 2020, assumed that the first high-volume full energy-assisted recording product supporting 24TB CMR capacities would ramp up sometime in 2022.

In order to mitigate risk in the case of further delays, HDD companies are undoubtedly investigating designs supporting more than 9 disks-per-drive, with some plans in place to eventually roll out a 10-disk design in the coming years.

In recent weeks, though, supply chain chatter has signaled that 10-platter designs may emerge sooner than originally planned, with some information pointing to 10-disk designs to support 20TB CMR capacities as a possibility. Granted, these plans could change, depending on pre-MAMR/HAMR areal density achievements, and technologies such as Western Digital’s EAMR approach may be employed to raise drive capacities over the next gen to obviate the need for a tenth disk.

Requests for substrate samples less than 0.5 mm in thickness are also materializing, with indications that drives with more than 10 disks are under investigation as potential future products.

If full energy-assisted recording yields less-than-expected areal density boosts or suffers continued delays, exploring the possibility of utilizing 11, 12 or even more disks in a 3.5″ HDD platform would be wise to do now. Design decisions such as substrate material choices as well as greater substrate, media and head production capacity requirements would fundamentally change investment trajectories for captive components as well as the supporting supply chain.

Despite the long history of HDDs, the industry should be commended on delivering a level of innovation and capability so critical to supporting the storage of exploding quantities of data, especially in the cloud. Public cloud infrastructure prior to the current pandemic already enabled countless consumer and corporate storage and compute services, scaling at rates far exceeding the past practices of corporations maintaining discrete on-premises data centers. With Covid-19, cloud companies were able to rapidly enable the seismic shifts to work-from-home and remote learning, with storage playing an important part in these changes.

While NAND flash and its associated SSDs garner so much attention for storage performance, no large cloud vendor today could envision a future that was not supported by vast quantities of HDDs that form the backbone all cloud infrastructure. HDD companies (all 3 of them) are the de facto stewards of the so-called datasphere.

While technology challenges continue to tax the capabilities of the HDD industry, all 3 players have been responsible for key enablers of high-capacity nearline HDDs over the past decade.

Toshiba was the first to launch a 9-platter HDD, Western Digital proved that helium-sealed nearline HDDs could provide a combination of higher capacities with more disks-per-drive while reducing operating power, generally leading the industry in capacity gains from 8TB up through 14TB. Seagate, now leading at 16TB, invested early in HAMR and has leveraged its consistency of direction to potentially lead the future transition to full energy-assisted recording.

Fundamentally, HDD companies will continue to pull out all the stops to drive HDD capacities higher in order to fuel the value proposition of nearline HDDs to address expanding cloud storage requirements.

However, future HDD design choices, the achievable capacity increments after 20TB CMR and the timing of the transition to full energy-assisted recording technologies all contain levels of uncertainty that will force the making of tough decisions over the next year or two.

The seeming elegance of future recording technologies may give way to more brute-force approaches to bridge widening time gaps to higher-capacity HDDs.