5 ms·
Seagate: 100TB HDDs Due in 2030, Multi-Actuator Drives to Become Common
- baal80spam 6y ago> Thank you for visiting this site. Unfortunately we have detected that you might be running custom adblocking scripts or installations that might interfere with the running of the site. This is a sure way to make me not visit a site ever again.
- nottorp 6y agoWorks fine with firefox/ublock origin, what else do you use? But speaking of idiocies like that, rockpapershotgun won't let me see subscription prices unless i agree to tracking cookies. Do they really think that's an incentive to subscribe?
- justinclift 6y agoBear in mind that Toms Hardware was also proven (a few months ao) to be selling their subscribers email addresses (to spammers?), even when they weren't supposed to.
- Exmoor 6y agoI'd appreciate more capacity, but at this point my big complaint is price per TB. I've been paying $15/TB for drives for, I think, 3-4 years. I'm beginning to wonder if we've hit the end of the line for decreasing storage costs.
- magicalhippo 6y agoSpeed is also getting to be an issue. Replacing a 14TB disk in a mirror can take almost a day (mine do ~220MB/s average). SATA is maximum of 6Gbps, but even maxing that (750MB/s) for a 100TB drive it would still take 37 hours to mirror a full disk.
- bcrl 6y agoHDDs will always have the constraint that they are dealing with media in 2 dimensions. Tape has the benefit that the magnetic media is thin enough that it can be rolled up inside the cartridge to take advantage of the 3rd dimension. Unless there is a significant breakthrough, HDDs in 10 years still won't be able to take advantage of the 3rd dimension. Flash has gone from SLC to MLC to TLC to QLC, and from 1 layer to 144+ layers. I think SSDs will pull even further ahead from HDDs over the next decade thanks to improved manufacturing techniques enabling more and more layers of nanometer scale features.
- ksec 6y agoThe problem is for years the platter density increase has been dismal. They were simply getting more platter for HDD capacity increase. I remember seagate talking about HAMR since early 2010s. And it is only now we got it working. It is sort of funny if we look at the tech how ~2014-2020ish seems to be the point where cost reduction in technology stopped or slowed down dramatically. Although I am quite optimistic that we will see $10/TB within this decade.
- AtlasBarfed 6y agoIt's almost like the market is a cartel with just a few players that don't really compete with each other, and are more likely to collude to increase or maintain prices like the DRAM makers.
- nottorp 6y agoI forget... is HAMR the slow one that kills your NAS?
- liminalsunset 6y agoI think you're thinking about SMR - Shingled Magnetic Recording. The tracks overlap because the heads are larger for writing than for reading, so writing can be a destructive process and data must be rewritten when adjacent data is written. These types of drives are fast for sequential writes, but any operations that edit data may be slow. Additionally, the drive spends time when idle clicking around to "defragment" the shingles IIRC.
- bcrl 6y agoSMR drives are not "fast" for writes. They run at slightly less than the same speed as non-SMR drives do for sequential writes (opening zones has a cost). They are unpredictable and often slow for everything else. SMR drives are somewhat like the first generation of SSD controllers: they lots of glass jaws that nobody ever expects when garbage collection begins to kick in.
- tinus_hn 6y ago> they lots of glass jaws that nobody ever expects Looks like autocorrect got involved here
- bcrl 6y agoI've been having pretty bad luck with autocarrot lately (yes, my phone has learned that word). The word "the" has come out as all kinds of weird things that barely resemble what I meant to type on my phone. It seems partially age related, as it seems to happen more often the older I get.
- magicalhippo 6y agoAs noted that's SMR[1], which is more or less orthogonal to HAMR[2]. That is, AFAIK, a drive can have HAMR without using SMR, or it can use both. [1]: https://en.wikipedia.org/wiki/Shingled_magnetic_recording https://en.wikipedia.org/wiki/Shingled_magnetic_recording [2]: https://en.wikipedia.org/wiki/Heat-assisted_magnetic_recording https://en.wikipedia.org/wiki/Heat-assisted_magnetic_recordi...
- nabla9 6y ago4 Gb/mm² for HDD density is a lot. For comparison: 7 Gb/mm2 - current NAND technology 0.3 Gb/mm² - current DRAM technology 0.2 GTr/mm² - current FinFET technology
- dogma1138 6y agoPure density isn’t everything, magnetic storage doesn’t have the same power and heat increases with storage density as solid state semiconductor storage has.
- 0-_-0 6y agoI'm waiting for the time when we will measure it in Gb/mm³
- AtlasBarfed 6y agoI didn't know NAND was that dense, is that for QLC 100+ layer? At this point it comes down to the expense of stamping out the NAND chips in order for SSD to pass HDDs in cost. Edit: https://www.eejournal.com/article/all-i-want-for-christmas-is-my-176-layer-flash/ https://www.eejournal.com/article/all-i-want-for-christmas-i... Good lord, Micron shipped 176 layer QLC/TLC compatible chips.
- bcrl 6y agoMeanwhile, 100TB SSDs already exist and can be purchased today. They're incredibly expensive at $40,000, but if you really need one... See https://nimbusdata.com/products/exadrive/pricing/ https://nimbusdata.com/products/exadrive/pricing/ for details.
- bcrl 6y agoWow, I'm getting downvoted for pointing out something relevant to compare against the article. It's relevant because what is a $40,000 SSD today is probably going to be 10-20 times cheaper in 2030 if trends over the past decade hold up. A 32GB SSD cost $85 in 2010. A $99 SSD in March 2021 gets you 1TB. It follows that in 2032, a 30TB SSD will be costing around $110. I don't see how a 100TB HDD will compete in that kind of market.