Tuesday, October 17, 2017

Will HAMR Happen?

For more than five years I've been skeptical of the storage industry's optimistic roadmaps in general, and the idea that HAMR (Heat Assisted Magnetic Recording) will replace the current PMR (Perpendicular Magnetic Recording) as the technology for hard disks any time soon. The first ship date for HAMR drives has been slipping in real time for nearly a decade, and last year Seagate slipped it again:
[Seagate] is targeting 2018 for HAMR drive deliveries, with a 16TB 3.5-inch drive planned, featuring 8 platters and 16 heads.
Now, Chris Mellor at The Register reports that:
WDC has given up on heat-assisted magnetic recording (HAMR) and is developing a microwave-assisted technique (MAMR) to push disk drive capacity up to 100TB by the 2030s.

It's able to do this with relatively incremental advances, avoiding the technological development barrier represented by HAMR. These developments include multi-stage head actuation and so-called Damascene head construction.
Below the fold, I assess this news.

Although HAMR was demonstrated in the lab back in 2012, making it work in production volumes is difficult:
But adding the laser-heating source to the read/write head adds cost and difficulty, and ensuring its reliability, longevity and also that of the recording medium as it gets intensely heated and cooled repeatedly is also a challenge.
A year and a half ago I wrote:
The disk vendors cannot raise prices significantly, doing so would accelerate the reduction in unit volume. Thus their income will decrease, and thus their ability to finance the investments needed to get HAMR and then BPM into the market. The longer they delay these investments, the more difficult it becomes to afford them. Thus it is possible that HAMR and likely that BPM will be "stranded technologies", advances we know how to build, but never actually deploy in volume.
It is starting to look like I was right. WDC's MAMR technology is different, and might be easier:
It adds microwaves to the write head, using a spin-torque oscillator (STO) to generate them. Electrons in a magnetised area have a spin state, tending to spin one way or another. By applying microwaves at the right frequency a resonance effect can alter the spin state and make it easier for the write head's electrical field to alter the magnetic polarity of the domain.
Perhaps the most interesting part of the report is that WDC:
thinks it can reach a 4Tbit/in2 areal density over time using MAMR technology, with a 15 per cent compound annual growth rate (CAGR) in capacity.
I've been writing abut the slowing of Kryder's Law from 30-40%/yr to 10-20%/yr since at least 2014, and that projection has been verified. Remembering that industry projections have a history of optimism, WDC's projection of a 15% Kryder rate going forward on the assumption that they can get a new recording technology into the market in 2020 should be treated skeptically. I would expect that the future Kryder rate will be more like 10% than 15%. I would also expect that the cost factor between disk and flash in the 2020s would be less than 10x rather than greater than 10x, but still more than enough to maintain hard disk's dominance of the bulk storage market.


David. said...

Seagate is still working on HAMR, and they just published a paper:

"Heat Assisted Magnetic Recording (HAMR) is the next generation hard disk drive technology that enables continued and significant areal density growth. ... In SMR, the tracks are written sequentially in bands with the tracks intentionally overlapping like shingles on a roof. In this paper, we introduce a novel track layout, Interlaced Magnetic Recording (IMR) and apply it to a HAMR recording system. With Heat Assisted Interlaced Magnetic Recording (HIMR), we observed a 31% increase in areal density over HAMR CMR whereas in a HAMR SMR architecture, we observed a 27% increase in areal density over HAMR CMR"

An extra 4% in areal density isn't a big deal, but IMR does allow random writes on 40% of the tracks, as opposed to SMR which enforces sequential writes on all tracks.

David. said...

Seagate continues to claim that HAMR is just one year away:

"Seagate claimed the new HDDs will debut to the market in 2018, and the company has plans to bring 20TB+ drives to market in 2019. If that isn't enough storage for you, the company plans to have 40TB+ models by 2023."

Note that Seagate has been working on HAMR for about two decades:

"Seagate has been working on heat assisted magnetic recording (HAMR) in one form or another since the late 1990s."