Comments

PixelStuff wrote on 3/29/2015, 8:39 PM
Hope it's not like the 60 TB hard drive prediction from 3 years ago.

http://www.extremetech.com/computing/122921-seagate-hits-1-terabit-per-square-inch-60tb-drives-on-their-way
ushere wrote on 3/29/2015, 9:00 PM
great, 10tb of data gone just like that ;-(

whilst i appreciate the need for ever greater total storage, i'm still wary of putting all my eggs in one basket...
Spectralis wrote on 3/29/2015, 10:40 PM
The whole Moore's Law thing has stalled. CPU development is crawling along as is HDD size. What has increased is the diversity of devices. Why have one product based on a certain chip when you can have 20? Rather than push the limits, manufacturers of chips and drives are making the most out of selling the same product in multiple devices for years before they develop the next version. The longer they can hold out and cut development costs by milking old tech the better for them. Now that Intel has no real competition it has no incentive to innovate. I doubt a commercial 10TB SSD will ever materialise based on current development. Even the GPU market is crawling along. NVidia has yet to release 8GB VRAM 980 cards which is pathetic considering the 7xx range had 4GB-6GB years ago. Even 8GB is just a measly 2GB increase in over two years. The 10TB drive will probably remain marketing hype with little substantial progress.

These companies will have to do something though because if they want to sell 4K equipment to the public with films taking up 200GB space plus the hardware necessary to edit and play them then the cost of SSD's will need to drop phenomenally and that's never going to happen. A 10TB SSD in 2016 will cost the price of a small car and there's no way that these companies can lower prices enough without significantly devaluing their current stock.
johnmeyer wrote on 3/29/2015, 11:18 PM
whilst i appreciate the need for ever greater total storage, i'm still wary of putting all my eggs in one basket...Did you feel the same way when disk drives were 10 megabytes? 100 megabytes? 1 terabyte? Why don't you have ten 100 MB drives???

My point is this: I don't think there is something magical about 1 terabyte that makes it "OK" to lose the data on it, but not on a 10 terabyte disk. Instead, I think you've just gotten used to the drive you've got now, just as you got used to the 10 MB, 100 MB, and 1 TB drive. In five years you'll be typing away at your computer that has a 10 TB drive, saying the same thing about the new 100 TB drive.

The whole Moore's Law thing has stalled. CPU development is crawling along as is HDD size. ... Rather than push the limits, manufacturers of chips and drives are making the most out of selling the same product in multiple devices for years before they develop the next version. The longer they can hold out and cut development costs by milking old tech the better for them. Now that Intel has no real competition it has no incentive to innovate. I doubt a commercial 10TB SSD will ever materialise based on current development..Well, I think your first statement is correct for CPU development, but the rest of what you say is a little off base.

Yes, clock speed maxed out about ten years ago. The laws of physics caught up with clock speed because the distance that electricity can travel between the metal traces on the chip becomes a factor. All the performance improvement since CPUs hit that 4 GHz "limit" has come from parallelism (mutliple cores).

Disk drives, however, continue to follow the learning curve (which is the underpinning of Moore's law,) and we continue to see remarkable progress, both in conventional drives, but even more so in SSDs.

Perhaps the reason you don't think that is happening with disk drives is due to a misunderstanding of Moore's law. For those who don't understand Moore's law, it is a variation of the famous "learning curve" which states that you get the same percentage increase (or decrease) in some variable every time the amount of product manufactured doubles. However, the fact that the "accumulated experience" (which is the phrase used to describe the total product manufactured by all companies) must double to get the same percentage change means that the market must continue to grow exponentially in order to keep getting the same percentage changes (in disk drive size) in the same amount of time (which is what Moore's law states).

However, if the market merely grows by a few percentage points, then you may not get the same percentage increase in disk size in the same amount of time. Gordon Moore tried to adapt the learning curve by using a proxy for doubling the output. He did this by stating that you'd get the same percentage increase every two years. It is amazing that this proxy worked as well as it did for as long as it did, but eventually it will fail to predict the future, even though the more correct version (the learning curve) is still able to make useful predictions.

So, you are right about the CPU, but I think you'll find that disk drives are still following the learning curve.

As for Intel having no real competition, and therefore no incentive to innovate, that simply isn't true. This reminds me of all the people who said that Microsoft owned the operating system and therefore didn't have any need to innovate. That was back when they were selling Windows 3.1 in the early 1990s. A quarter century later, we've had Windows 3.11, Windows ME, Windows 2000, XP 7, 8 , & 9. More important, far from having no competition, Linux killed them in the server business.

As for Intel having no competition in the SSD business, they actually aren't even the market leader. According to a December 2014 article in Forbes, Samsung leads the market with a share of 28%. Intel was second with 13.7%. That is a relative market share of over 2:1. For those not familiar with marketing, relative market share is a very importent competitive metric, and when you have a relative market share where the leader has a 2:1 lead over its 2nd place competitor, that leader is said to be dominant. So, in this case, Intel is fighting for its life in this market. In fact, according to this article, Intel is losing market share to both Toshiba and Sandisk, and may very well be in fourth place at this point in time.

Look At The SSD Market: Where Do Seagate and Western Digital Stand?

john_dennis wrote on 3/29/2015, 11:35 PM
Rather than respond to the [I]development[/I] side of the discussion, I'll what-if on the [I]application or product marketing[/I] side.

The last spinning disk I bought cost about $40 per TB whie the current price of Samsung EVOs in 250GB units runs about $400 per TB. I'd drive to Frys tomorrow and buy a 10 TB SSD in a 3.5" form factor for $80 per TB or $800. That's where I think I am on the of demand curve for bulk storage.

Based on costs and yields, the manufacturers probably don't see me and my kind as a legitimate market.
johnmeyer wrote on 3/29/2015, 11:58 PM
The demand for SSDs is going to be through the roof. These gigantic server farms (from Google on down) are growing at unfathomable rates, with no end in sight. SSD promises to cuts size requirements substantially, and it appears that they may consume less power, although I haven't seen any power per terabyte comparisons. If a server farm can cut both its power consumption and space requirements by two-thirds, that is going to drive a lot of demand which, in turn, will drive the prices down even faster than they are already falling.

While not an SSD, I did just get a 32 GB thumb drive for $11. That is still 5-8x more expensive than a really fast hard drive, it is about 1/3 what I was paying just 18 months ago.
Spectralis wrote on 3/30/2015, 8:38 PM
Here in the UK a 1TB SSD is £300-£400 which is $450-$600 so when commercial 10TB disks are released in a years time (don't hold your breath) it'll be a miracle if the ordinary consumer can afford them - even 5 years later. The prices of 4TB HDD's are just about reaching the cost of 3TB drives 2 years ago so based on current price patterns, drive manufacturers are never going to cut their own throats just to throw us a bone. The first 1TB drive was released in 2007 and eight years later we're only just getting 6TB drives so if we go by capacity then they don't follow Moore's Law.

Moore's Law was a conjecture about processor performance that has been revised at least twice and has never been historically accurate anyway. I quoted it in the context of price vs performance because manufactures like to sell the media this cyberutopian future but the reality is flogging old tech for as long as possible to ramp up profits. Even though I'm not Mystic Meg, I can safely predict that 10TB drives will not cost $800/£550 any time soon.
Spectralis wrote on 3/30/2015, 8:51 PM
Concerning Intel I was referring to their CPU development for which they are famous and hold a virtual monopoly. I don't think many consumers know or care about their position regarding SSD development. The point is that SSD development in relation to falling prices is little different from HDD's because manufacturers are compelled to meet development costs and that means selling old tech for as long as possible. They have an unspoken cartel that none of them wants to break.
PixelStuff wrote on 3/31/2015, 12:00 AM
I think Intel has actually been getting competition from the low voltage mobile market recently so they have had to split their focus between all out performance and saving as much electricity as possible. I also think they are bumping into the laws of physics on the high end so they (and their competition) are having a harder time there also.

Last year we kept reading about the 14nm process getting delayed, and I don't think it was purely because Intel has minimal competition.

ushere wrote on 3/31/2015, 5:46 AM
[i]whilst i appreciate the need for ever greater total storage, i'm still wary of putting all my eggs in one basket...
Did you feel the same way when disk drives were 10 megabytes? 100 megabytes? 1 terabyte? Why don't you have ten 100 MB drives???

My point is this: I don't think there is something magical about 1 terabyte that makes it "OK" to lose the data on it, but not on a 10 terabyte disk. Instead, I think you've just gotten used to the drive you've got now, just as you got used to the 10 MB, 100 MB, and 1 TB drive. In five years you'll be typing away at your computer that has a 10 TB drive, saying the same thing about the new 100 TB drive.[i/]

sorry john, nothing is going to make me store all my client projects on a 10tb (or bigger) drive just because i can. i will always break up my storage into 'manageable' sizes and in the case of client projects, separate hd's.

i'm not a luddite, but i think that until the reliability and speed of transfer of large capacity (external) hd's is WELL established i have no desire to be the first to say bu**er when i lose 10tb of data*.

*and of course, i'd need another 10tb hd for back up ;-)
JJKizak wrote on 3/31/2015, 6:24 AM
Spectrails:
Check the website out called "Optalysys" and you might change your mind about Moore's Law stalling. Intel will be history in a few months unless they buy this company out.
JJK
Chienworks wrote on 3/31/2015, 8:32 AM
I always stay at about half the size of the bleeding edge of storage. I am just this year putting in 2TB drives in my systems as main storage. Now that 4/6TB drives are out the manufacturers have had enough time to weed out the problems with producing 2TB. Failure rates are minimal and complaints are nearly gone. In 3 or 4 years when these 2TB drives start wearing out i'll look into whatever is stable then.

Of course, with mirrors and backups, i'm not as concerned about any particular drive going bad.
Larry Clifford wrote on 4/3/2015, 3:37 PM
Johnmeyer wrote: While not an SSD, I did just get a 32 GB thumb drive for $11. That is still 5-8x more expensive than a really fast hard drive, it is about 1/3 what I was paying just 18 months ago.

Will you provide a link where you bought this?
Are you satisfied with it?

Thank you.
johnmeyer wrote on 4/3/2015, 4:01 PM
Will you provide a link where you bought this?I got it at a Costco store. It is available online for $2 more, although they appear to be temporarily out of stock:

SanDisk 32 GB Ultra 3.0 USB Flash Drive

It is a Sandisk UBS 3.0 Ultra, Item # 948032, UPC 619659107123, Model #: SDCZ48-032G-AC46.

You can search on the model # and will find it on Amazon, although the specs there say USB 2.0 and the transfer speed doesn't match what I found on my packaging.

I am using these to send out HD videos, and so as long as the speed is fast enough to play HD (which it is), it is good enough for my application.

I haven't actually tried to benchmark it.

Is there benchmarking software for thumb drives ??
Dach wrote on 4/4/2015, 8:24 AM
Just for kicks, I still shake my head amazed with advances in storage devices. Especially a flash drive considering once upon a time I used to have boxes of floppy drives.

Chienworks wrote on 4/4/2015, 5:55 PM
I had just posted on facebook yesterday the observation that back not that long ago we used to back up our hard drives to CD-R, but it wouldn't be feasible now as it would take over 13,400 CD-Rs for me to back up my main editing station.

At my previous job we had a shelf full of floppies, about 450 of them. I was able to copy the entire batch over to one single CD-R. Now my phone has enough memory in it to store the equivalent of 92 CDs.
johnmeyer wrote on 4/4/2015, 8:34 PM
Ah yes, backup to floppies. I remember Fastback. I used it with 360k floppies, then thought it was heaven when I could use it with 1.2 M floppies, and eventually 1.44 MB.

As for the number of CD-Rs required, I think some who have posted in this thread may still be using 100 MB disk drives, since they're worried about having a drive that is too large. For them, CD-Rs may still be viable. DVDs for sure.