I think that the market for really large capacity music boxes is pretty limited. From what I have seen only Zune and Ipod have anything over 32 gig. Creative only goes up to 32 gig. There just aren’t that many people with big music collections like you or me.
No, I dont think so. Thats just one failure mode. Flash drives, like any piece of electronics, has several. Failure of the flash media is another. I have usb drives with a bad piece of flash on them. It happens. Failure of the supporting circuitry on the flash board is another.
The one failure everyone has mentioned is just the normal process of blocks going bad on flash media because of overwrite. The firmware of the flash drive will do its best to shuffle things around, but as the overwrite limit gets reached, it will not be able to write anymore. That’s just normal aging in the flash world. That doesnt mean that the thing wont die from other causes.
I disagree with this too. Most drives die slowly. We can see bad sectors add up and bad read/writes. Most people dont check their smart settings or bother to do anything when chkdsk starts reporting bad blocks. I think its rare for a drive to just stop working. Of all the failures Ive had both in home and work, I cant remember one that just died without warning outside of a RAID array. Typically you would start seeing signs or corruption: OS crashes, application crashes, clicking, etc.
I think the marketing and the hype of SSD drives is pretty out there. Im skeptical that all failures will leave you with a readable disk and all the other myths out there.
Y’all do realize that solid state storage is already the norm on the majority of the computers today - except for desktops and laptops?
-
Cite?
-
You left out servers, which by and large do not use SSD drives.
Moore’s law is often described as a doubling of the number of transistors per dollar. Whether or not you agree with that particular definition, it holds as well as the others.
Of course HDD prices and capacities follow Kryder’s Law, which tends to follow a pretty similar curve to Moore’s.
So, as everyone has pointed out, we’re not likely to see SSD’s cheaper than HDDs in the near future, at least not at the top end. But that’s not really what the OP cares about. The OP cares whether they will become cheap enough that they are overwhelmingly used in personal computers. And that’s definitely happening. Note that they don’t actually have to become cheaper than HDDs, since they have other advantages, like no moving parts, faster seek times (with good firmware), more graceful failure, and (potentially) smaller footprints.
I’m assuming that the OP is looking for a laptop. If he’s getting a desktop, it’ll take longer (and might not actually happen, since desktops are not subject to the same same physical strains that make SSDs a good choice for laptops).
A growing trend in laptops has to grow cheaper and smaller, rather than more powerful. Effectively, the market has decided that memory size and processing speed are “good enough”, and other things are more important. You can already get netbooks and other laptops with SSDs for a premium. As time goes on, that premium will shrink for a given size. I’m not going to make some silly claim like “120GB storage will be enough forever”, but it will be enough for a few years, for many people. Right now, there’s a $200 premium on having that 120GB be an SSD. In a year or two, it’ll be half that, and another year, half that again, but 120GB will still be enough for many people, making the small premium well worth it for the other advantages.
Oh yeah, just as an aside, when ruling out desktops and laptops, what exactly is left when referring to the majority of computers? We can leave out servers, I covered that in my last post. The next most popular product after desktops and laptops is probably game consoles, those don’t use SSD either.
The hard drive on my laptop just died - no warnings at all.
I spend a lot of time looking at field failures, and anything mechanical fails at a far higher rate than anything purely digital. Bad bits in either can be repaired silently of course. The can be board fails for both ( but that shouldn’t result in data loss since you can swap chips.) I can see truly catastrophic fails from bad I/O buffers or address decode logic, but that is a tiny portion of the chip. Very complex microprocessors manufactured using aggressive process technologies are several orders of magnitude more reliable than hard drives according to the data I have. We don’t use solid state memories, so I don’t have any decent data on them. I suspect that when there is a market for them as real hard drives they can be made a lot more reliable (and a bit more expensive) than the ones being given away now as thumb drives.
Embedded systems, which are everywhere and rarely use traditional HDs.
Magnetic drives may fail by sector rather than all at once, but the time at which they fail is entirely unknown. Even if you put a flash file system on a magnetic disk drive, it wouldn’t be able to work around the gaps because bad sectors aren’t caused by data writes. With a flash drive, the time of failure is always when you update a sector with new data, so you can immediately check for errors while you still have the original data handy.
If a data sector on a flash drive can go bad at any unknown point of time, I haven’t heard of it, but I suspect that it’s longer than the time it takes for you to come to its limit of write cycles.
You are correct that “by and large” data centers don’t use SSDs, but they are there, and it’s a fast growing market. The reduced power and heat, in addition to the dramatic increase in read access, makes them a great alternative to HDD arrays. And, of course, enterprise data is generally well backed up to mitigate the shorter lifespan of SSDs.
Enterprise benefits are SSD are non-existant, at least for now.
Again, the net is full of SSD hype. To address the original question: no, SSD will not take over anytime soon. Perhaps we’ll see them more in high-end laptops, particularly if Apple starts doing it and HP and the rest follow suit.
Since, unlike an optical drive, SSDs don’t have to be one large contiguous device, could we see a time when a large SSD drive was really just a collection of smaller ones?
Compare an optical drive to a paper scroll. Once the size/shape of the scroll is set, you can’t change it. If you accidentally destroy a section in the middle, then it’s no longer one scroll. Hard drives are kind of like that.
Now what if we had an SSD drive more like a 3-ring binder. Lots of pages of stuff. If a page gets damaged, you can open the rings, and replace that page. Yes, you might have to copy (or lose) the contents, but it’s just of that page. In addition, if your original 100 pages aren’t enough, you open the rings up and add another 100.
Could an SSD drive be created that way? So if I want a bigger drive, rather then taking out the old 300G and buying a new 512G, I just buy and additional 200G and add it to what I already have. And if one part of it starts to accumulate bad blocks, you could move the rest of the data from that section, pull it and replace it.
I think that this is already being done - also with magnetic hard drives, really. It’s a bit hard to start with one drive and ‘expand’ it up to a multiple drive system, but you have RAID arrays and similar setups where multiple physical hard drive units appear as one drive.
I believe that some netbooks also have an ‘expansion slot’ option available - if you start with a 10 gig SSD, and add an 8 gig SD card in the expansion slot, it appears as an 18 gig C:\ drive.
I’m not sure if any of that is then usable if you take out/lose the SD card.
And yet, people are buying and/or selling them.
A quick peek the links searching for “SSD” on theregister.co.uk finds the following enterprise SSD-related articles, going back as far as June:
Some shops, at least, seem to feel that they’re useful, Microsoft’s opinion notwithstanding.
I think it will take a couple of years.
They need to be bigger, and cheaper.
I have SSD’s as my OS drives in my laptop and my desktop.
I don’t think I could go back. (Just need them to be bigger.)
Hard drives actually fail in a very similar way to SSDs. Blocks steadily go bad one by one, they are replaced from a pool of spares (error-correcting codes recover the data in failed blocks, so the process is transparent). Once the spares are gone, you may throw the whole thing out. Monitoring your SMART information is important in knowing how far along your drives are. Hard drives, however, are much more likely to suffer from instant catastrophic failure or some other deviation of normal decay.
PatriotGrrrl must have been referring to embedded applications, which do indeed constitute “most computers.” Those always use flash, unless they need a lot of capacity.
Hm, I didn’t know that. I thought that sort of process was unique to SSDs. I guess not.
I haven’t looked deep into HD drivers recently, but years ago, in the microcomputer market, only Novell, not Microsoft, was doing that kind of real-time repair after the initial drive format. Are you sure that is built in to today’s Windows and/or Mac OPSYS?
Can someone comment on this?
Are we talking picoseconds, nanoseconds, milliseconds … a vastly noticeable boot time?
Given the price of regular drive storage, partitioning off a few feet for emergency boot backup would cover failures (just change BIOS settings and remember to backup settings should do the trick). In a bit of time, maybe size will be cheap enough to have OS and programs on there, with magnetic as data storage.
One device that I worked on was a full Linux install on a PDA sort of device, where the OS lived in flash memory. I’d say that full bootup was probably around 5 seconds from the time you hit the power button. Not instantaneous by any means, but it was definitely a lot faster startup than turning on a PC.