Anyone who still believes that CD, DVD or Blue-Ray still have a future is deluding themselves. There is no market for spinning media, it's just that there is currently no better alternative. Flash Memory will replace all spinning media as soon as it becomes viable from a cost perspective.
FYI: Blue ray is up to 400GB/disk (As in you can make 400GB disks that a PS3 will read) and it costs a tiny fraction of what that what flash memory costs. In ~20 years they might cost the same per GB, but that's still a long way from now and plenty of time for the next optical disk format to show up.
PS: I am assuming blue ray disks will cost around 10c per disk in 20 years. So flash needs to drop to 1/10,000th of it's current price to be competitive in that time scale.
The point is not how much you can store and how fast you can access consecutive sectors. The problem is that spinning platters have been stuck at 100 random seeks / second for the past 20 years, and most database struggles have been based on that. The new Intel SSDs can do 100,000 "disk seeks" per second, which means that you suddenly have the speed of a thousand-disk raid on one SSD. That's why SSDs are interesting. Not the storage per disk, the random-access IO. People still use tape (not just .tar formats) because it's really cheap for dumping big backups.
But I don't care about random-access speed when I'm just using it for media (music, photos, video).
I think within a few years most people will end up with a small (a few hundred GB) SSD for their OS and programs, and a very large traditional HD (a few TB) for archiving data that doesn't need random access.
That's the system I'm using now. Two 73GB 15K drives in Raid-1 for my OS, with 3 1TB drives in Raid-z for media.
Yeah, the same separation between small assets available quickly and big assets available slowly is how Amazon recommends you separate data between SimpleDB and S3.
I'm quite fascinated by all the ways we'll be able to persist data.
Yep, it isn't a new idea. The premise of Von Neumann machines is that you work around the cpu/data bottleneck with increasingly faster (and larger) caches. You have a tiny bit of ludicrously fast on chip memory (registers), a few megs of successively larger and slower L1/L2/L3 cache, a few gigs of kind of slow memory, and hundreds of gigs of unbelievably slow hard disk. This would just split the unbelievably slow disk into two separate layers - kind of slow SSD and really slow disk.
Yeah, I've heard the argument that the blu-ray/hd-dvd format war was kind of moot because digital distribution is the future, but we are still a few years away from being able to download/stream the same quality that you get out of blu-ray. Any HD content you can get online at the moment is heavily compressed and often "HD" in resolution only.
I would love music, movie to be sold on SSD, but the high fixed cost ($2) is unlikely to change much, and it really eats into margins on low-priced products (eg a $20 DVD). And what about online delivery?
For what it's worth, chipmakers been saying that flash drives will catch up with hard drives in price for about 17 years now. I first read an article about solid state drives catching up with rotating drives in 1992.
Right now a terabyte hard drive costs $81, and a terabyte of SSD costs $2684, I'm not going to hold my breath.
They're not really saying that SSD will match HDD price per GB, but that it will reach parity for a specific demanded GB capacity (64-256GB) in a few years.
But by then, as you imply, the demanded capacity also will have increased...
So it seems to be a question of the rate differential. So far, SSD's are getting cheaper at a much faster rate than demanded capacities are increasing, and (assuming nothing changes the rates, like Blu-ray-sized 400GB movie downloads), it will catch up soon. Firstly just nipping at the heels of the low end, then biting a chunk out of the mainstream, and eventually consuming all but niche applications.
It has already caught up for lower capacities e.g., a 1 GB flash drive is cheaper than a 1 GB HDD (fixed costs: $2 vs $40-50), and disrupted hard drives from iPods and netbooks.
Meanwhile, desktops PCs themselves are being disrupted by laptops, which are being disrupted by netbooks.
The lower capacities have become cheaper in SSD, however that's usually because hard drives rarely get too small. A 3.5" is always roughly double a 2.5", so if the largest 3.5" disk is currently 1TB, then the biggest selling 2.5" is likely to be a 1/2TB, which are the ones I generally look at.
However, I've already been looking into the SDD vs HDD and I'm swaying towards the HDD. I've already seen multiple USB flash failures, thankfully none were carrying data I didn't have backed up, however in a couple of situations we've got lucky; my wife's 8gb flash drive decided to fail when she needed it for college, so I had to load the back-up of her entire disk onto my FTP.
The USB flash cost $20 for 8gb, which was an introductory price as it's now $30 again. It had a 6 month warranty and failed a little bit past that line, the other one she got at the same time is running fine so I'm not going to call it as purposeful failures, but good warranty placing. However, for 4 times the price we paid, AKA $80, I've seen 160 gig 2.5" HDD with a warranty of 5 fucking years! Assuming flash drive sizes double every year for a similar base price of $20 then it's going to take at least 4 1/2 years for SSD to match the HDD and I'd have complete warranty on the HDD!
So IMO for portable storage, I'm not touching the piece of cheap crap flash drives. I'll stick with real technology, with real warranties that haven't failed me once since before the 3.5" floppy disk was released!
SSD's are getting cheaper at a much faster rate than demanded capacities are increasing ... Firstly just nipping at the heels of the low end, then biting a chunk out of the mainstream, and eventually consuming all but niche applications
This is a classic Innovator's Dilemma/Innovator's Solution situation. People were buying for capacity, soon they'll be buying for convenience (speed, power consumption, etc). Tape drives are cheaper than HDD but no one is buying those.
The SSD taking over the HDD is based on the preposition that everything stays the same, but the game is going to change and we all know that. Digital distribution is beginning to hit big, I mean look at NetFlix on Xbox 360 and iTunes.
I haven't bought a CD in over a year and when I did it was because it was one of my favorite bands. How long is it before I'm saying the same thing with DVD?
SSD might pick up some niche uses, but there's a reason people have wanted mass storage, which is that programs are progressively getting bigger and so are all the files we store. I remember thinking a 400x300 pixel image was big, now I think a photo smaller than like 2000x1500 as small. In 5-10 years I'm going to think anything smaller than like 10,000x7,500 as pathetically small. Equally, I'm going to think any video that's not at least 720p isn't worth having, because by 2015 companies want UHDV to be released displaying at 4320p. When my video is taking up 32MB per frame I don't want, I'll need maximum capacity still.
Ok, that's 75 megapixels. Ken Rockwell estimates that 35mm film is about 175 mp (http://www.kenrockwell.com/tech/film-resolution.htm), so that is a technically realistic number. But you need an outstanding camera and photographer to make all those megapixels matter, and most people won't pay for and won't even notice the extra detail. Resolution bigger than the biggest computer doesn't matter much for mainstream users, and even a 6.1mp image from my DSLR looks good printed out at 8x10".
Now a hard drive full of HD movies would take up tons of space, but do you want to carry that around with you? Not to mention that movie studios will do whatever than can to prevent persistent files from being the digital format of choice. They would much rather stream on demand and charge per viewing or per rental.
We've always wanted more space, but computers are getting to the point where they're fast enough and big enough to satisfy common users. The fact that you're here on HN means that you're by definition not a common user.
What I find most attractive in SSD is the IO performance.
The fusion-io PCI express SSD storage is the most attractive. The IO speed difference and reduced latency would make a significant difference for my application.
I'm not so sure about long term storage reliability. I would currently prefer to combine both type of storage.
I couldn't resist the hype and bought the cheapest no-name (A-Data, anyone?) 32GB second-hand SSD. I haven't seen a noticable performance improvement since but FWIW the silence - getting rid of the disk thrashing - is worth alone the price of this thing.
I too bought a 32GB SSD from SanDisk a year ago to replace hard drive in my IBM ThinkPad X60s to get silence and shock resistance (I guess I'm paranoid on those issues, but it made me happy) Sequential write perfomance is noticeably less than even in 5400 rpm drives, but I didn't care much; it's okay for me.
Yet I would call people who use SSD in their computers early adopters; the outlook for SSDs is optimistic but I guess it will be a few years at best till SSDs will be able to become mainstream.
What would be really informative is a graph of the historical price per memory size for both spinning and solid-state harddrives... anyone come across anything like that?
