I Don't Think So,
There is, in my opinion,
no way it will take until 2021 to reach 2TB flash drives, I do understand you have put more thought into the subject than I just have but I reckon they'll arrive around 2015 at the latest...
SanDisk and Sony have announced a roadmap for Memory Stick products that take it from its current (Memory Stick PRO) 16GB capacity up to 2TB. Do we have a tape-like generational roadmap here? LTO is a classic example of tape generations with LTO 4 at 800GB native capacity being the current generation and LTOs 5 and 6 ahead, if …
I'm not going to dispute the figures like other people have done, but the article makes a single obvious mistake that has afflicted techies (including me,) since the invention of the wheel.
You've forgotten to include in your figures that no-one will still be making rotating drives by 2013, due to no-one being able to profitably make them.
As their market shrinks, their profit margin per unit will need to go up to compensate. This can happen in one of two ways.
Either they'll have to charge more for them (encouraging the decline in sales relative to SSD,) or they'll have to invest less in development (encouraging the decline in sales relative to SSD.)
By 2013 the only people still buying rotating media will be those with ludicrously high storage / retrieval compromises, such as god knows who.
As density increases, number of electrons storing state decreases dramatically. Those electrons have a tendency to leak, weakening the state over the time. Unless SSD manufacturers develop some new materials, we will hit a wall sooner than later. Quite similar with HDD, although for different reasons. I don't believe we will get to 2TB without some breakthroughs.
Tut tut, fall asleep in school when they covered compound interest did we? ;)
Double your 2021 figures, and that's round about what you get for a 50% increase per year (which, incidentally, I think is an underestimate). So, 4TB memory sticks and 64TB SSDs and HDs all around by 2021. Or not much later than 2015 if you ask me.
It's also interesting that you felt the need to plug the exact same figure into the exact same formula twice (SSD & HD) to work out that we get parity. :)
The SSD prediction is totally off as a 1TB 2.5 drive has been announced at http://www.marketwire.com/press-release/Puresilicon-936099.html
The other predictions are also off as Sony release a new Memory Stick every year with double the capacity of last year and hard drive manufacturers are going to have to increase capacity to keep up with SSDs.
What you write makes sense - but people have been writing obituaries for rotating media since the late '80s, and rotating media has yet to get the message and die.
It's certainly instinct to look at margins and R&D pressure and say, "This is untenable" - I'd rather open a "Curves For Women" fitness club in Saudi Arabia than get into the hard drive business - but things have looked that way for a while now. And yet here we are.
That said, prognosticating about the future of technology in ANY sense isn't so much a minefield as a land mass which is actually entirely composed of mines, with little bits of dirt scattered about. If it weren't, we'd all be using thin client set-top-boxes with bernoulli drives.
"A second by-product of increased bit density is reduced write endurance. This is being countered by adding parallel channels to controllers and avoiding unnecessary erase-delete cycles when writing data by over-provisioning the NAND and by clever file systems such as SanDisk's ExtremeFFS. This is a uniquely flash phenomenon; it doesn't apply to tape or to hard drives."
sorry, but until these problems are overcome, not just to some degree 'countered', I think it would be foolish for those concerned about the integrity of thier data to use flash memory in a computer system as primary storage / use
it's like expecting bad sectors in HDDs and just not using those sectors over time .. you replace an HDD when bad sectors start showing, you don't just 'counter' a problem with your storage medium degrading like this
I still boot up an old 200MMX machine with a 2.8GB drive that I keep some old files on I occasionally need .. and just for the fun of seeing my first PC still running .. ran SpinRite on it in 2007 .. no problems at all
until they solve the write speed and re-write durability problems, I think HDDs will be around for many years
and as far as capacity , most users today will never fill even 80GB in a consumer PC .. I'd take a stable proven 120GB HDD over a 250GB > 500GB in a new machine .. not having a FAIL is much more important than HDD size bragging rights
Even the guess of a doubling per 2 years doesn't match reality. I've been watching the developments vaguely since 2005, and seriously in the last 18 months. A 1G USB stick cost me eur80 in 2005, and I can now buy 32G for eur60, and 64G for eur90. So at constant price, there has been a doubling in size roughly every 6 months.
I read an article in the 80's that predicted parity for HDs and SSDs in 2015, and it looks like they got it just about right (they didn't express it that way, but that is how you would interpret it now).
Graph of recent USB stick prices here (search for "USB"):
Most 'Home Users' these days like to save the odd UTube clip, and more than the occasional Music Video on their drive, and pretty much never delete anything. 80GB is nothing - look how many iPod's etc. come with more than that - it is only the lightweight flash-based ones that have less than 40GB these days.
It's misleading to say that the write duty cycle doesn't apply to tape drive. LTO4 has a (complete) read/write durability of 200 end-to-end passes (at least from one manufacturer). That makes it suitable only for backup and archival purposes. Well, OK - ti applies to writes and reads and it is the medium, not the drive that has the limit. Written once per week and an LTO4 tape will only last 4 years.
Tape drives will, of course, potentially last longer, but only if they are regularly maintained (especially if the duty cycle is high).
As for Erik Aamot and don't use flash SSD for primary storage, that's nonsense. All drives are prone to failure, whether flash or physical. They also wear out (and people shouldn't confuse MTBF with expected lifetime of a drive - the latter is a lot less than the former these days). It's for exactly those reasons that RAID is used along with other techniques to maintain data integrity. We have 10s of thousands of disk drives with Petabyes of storage and modern arrays allow for the hot replacement of failing devices without service disruption. The practical issues of write duty cycles on SLC flash are frankly not a problem with proper controller logic.
It might well be that 3 & 4 bit MLC does not yet have enough write duty cycles for use on write-intensive systems, but SLC (albeit at a cost) is most certainly suitable now.
The only advantage that HDD has over the best SLC flash is cost (albeit that's a very large issue). On performance SLC massacres HDD and in 3-4 years time top-end hard drives will be obsolete. HDDs will then be handling the bulk access, high-capacity, mdium access speed market.
I'm surprised that there wasn't any mention of the mechanical hard disk's Achilles heel: seek time. It takes at least 5mS (usually more like twice that) for a hard disk to change the cylinder address on the disk from or to which it will next read. In contrast, "seek" for a SSD is a matter of microseconds, if it means anything at all in that context.
It's my opinion that in the shorter term (i.e. the next decade) the most fruitful developments will be filestores and database stores that combine a SSD and a hard disk, so that the large chunks of data that are accessed sequentially reside on the (cheaper) hard disk, and the smaller, heavily and randomly accessed chunks reside on the SSD.
It can be done at the controller level, to avoid any need to introduce new system software. Put a few hundred Mb of fast flash memory on the circuit board of a hard disk storing a few hundred Gb, and let the disk controller use it as a cache. Unlike RAM cache, there is no problem if the disk drive loses power.
In short, it's not simple competition. These two technologies should be synergetic, at least until solid-state costs little more per Mbyte than hard disk.
A disk drive takes around ten watts to keep spinning. At 10p per kW/h, 24x365 operation for 3 years, that's £26, something like a 25% to 50% overhead on the price of the disk.
Anyone know how the energy cost of maintaining solid-state storage compares? I imagine that at some storage density, it becomes the greener technology, giving it a growing life-of-product cost advantage. Indeed, ultimately might we see rotating disks banned in the same way that light bulbs have been?
Biting the hand that feeds IT © 1998–2018