Slashdot Mirror
SSD Prices Down 46% Since 2011
crookedvulture writes "Hard drive prices have yet to return to normal after last year's Thailand flooding. There's good news on the solid-state front, though. The current generation of SSDs has steadily become much cheaper over the last year or so. SSD prices have dropped an average of 46% since early 2011. Intel has largely shied away from discounting its drives, but the aggressive competition between other players in the market seems to have forced its hand. There's no indication that competition is waning, suggesting the downward trend will continue. Right now, an impressive number of drives are available for less than a dollar per gigabyte."
SSD means Solid State Disc, a faster permanent storage type than HDD, however lacking capacity-per-dollar of HDDs.
Solid State Drive *
Salsa Saturated Dorito
I work for the Department of Redundancy Department.
SSDs also have OCZ and Crucial leveraging MLC and SandForce's controllers to deliver optimized and boosted performance and extended life for reduced cost. SandForce SF-2200 chipsets compress data as it goes out to the chips, reducing write volume and thus giving fractional write amplification. This improves performance and reduces storage wear, improving product lifetime--hence the use of MLC. Of course already compressed data doesn't have those benefits, hence why OCZ's Vertex line has better write speeds--they use synchronous chips that write as fast as they read (Agility drives use much cheaper chips that read faster than they write, so for compressible data they're FAST but for non-compressible data they're slow), and use compression just to extend drive lifetime.
With all the manufacturers making good use of SandForce's better chips, and SandForce's strategic pricing (read: they're relatively cheap because they want to be a major consumer and enterprise supplier of SSD controllers, which would make them richer than charging a fistful of cash per chip), a lot of inexpensive SSDs have shown up. Essentially Intel tried to hold prices high, and SandForce stepped up and decided to help the whole market undercut them in order to gain market dominance (Intel uses SF chips in 2 models; they previously used Intel proprietary controllers, and have also used Marvell controllers).
That's called "competition," son. It's what big businesses try to prevent with patents, lock-in, vertical integration (so you can't undercut their prices ever), supply chain control (so you can't get the raw materials to make a competing product without buying from them), etc.
Support my political activism on Patreon.
Recently we put some higher-end drives in our servers and we love the hell out of them.
There have been a couple of reports that just came out that suggest that SSDs can pay for themselves in many server environments. It isn't just the power savings (remember that non-residential customers often get charged more per kWh), but that many servers are often disk I/O bound. When you replace spinning platter drives with SSDs, you might be able to cut your server count in half. Admins have found that they can completely eliminate caching web servers because the app servers can crank out so much.
Only if it's Win 7 32-bit. If he's using the 64-bit version (which is a good bet) he'll get the full 8.
Paranoid much? I post that link because the guys at coding horror see no problem with plopping down a grand on a drive and like /. is full of geeks that know what they are doing so you can't blame it on noob mistakes.
Yes you can. I've seen Jeff Atwood be very clueless and noobish indeed. He's a programmer, more of a web developer than anything else, not an internationally recognized authority on SSDs. He just likes them because they're fast, and he (and his buddies) had a bad string of luck with them. Every time you link to that one blogpost on codinghorror, you're linking to one guy's anecdote, nothing more. He doesn't give any technical reason to justify his opinion that SSDs are inherently failure prone. He just says he and his friend had a lot of failures, conflates that into a general problem, and uses a stupid sexist analogy to cement the idea that SSDs are fast but inherently unreliable into the skulls of idiots like you.
i would also point out that article isn't even a year and a half old so if you can provide your OWN link showing this miracle breakthrough that has eliminated SSD failures I'm sure the guys here would be happy to read it.
The 'miracle breakthrough' is neither a miracle nor a breakthrough, it's just hard work. SSD reliability depends a great deal on SSD firmware, because managing flash storage is complex. Lots of SSD firmware was written with more of an eye towards time-to-market (companies trying to carve out marketshare early) than wringing out all the bugs before they shipped.
If you want a reliable SSD, stick with the vendors who have managed to sell drives to major OEMs. For instance, if you can get a drive closely related to any of the ones Apple OEMs for MacBooks, you're probably in good shape because Apple does extremely strict qualification (acceptance) testing on all storage devices they ship.
If you go down to the comments you will see failure after failure, every major brand and model, and these guys again do NOT go cheap so you can use the CCC (Cheapo Chinese Crap) excuse either.
Many of the failures reported in Atwood's initial blogpost (and lots of them from the comments) were brands like OCZ and other "enthusiast" overclocker kiddie brands. OCZ in particular is well known for its cavalier attitude about treating customers as SSD firmware beta testers.
We are talking top o' the line drives by reputable companies like Intel crapping all over themselves.
Just curious, did you mean the comments like this one?
"Over at blekko, we've had 3 SSD failures after 1.5 years, out of 700 drives. These are Intel X-25M 160G2 drives." - Greg Lindahl
That's an actual statistically significant sample, not just an anecdote! And it's a very low failure rate; HDD vendors would probably kill for 0.43% over 3 years.
And then there's the comment with a link to Anandtech which gives some more real data, i.e. return rates recorded by a large French etailer for several major brands:
http://www.anandtech.com/show/4202/the-intel-ssd-510-review/3
(the worst is OCZ at almost 3% -- so even keeping in mind OCZ's questionable practices, the picture is quite a bit different from what you're trying to paint, eh?)
And there's also a ton of anecdotal comments which say "I have a SSD and I never had a hint of trouble with it, shrug", which actually seem to outweigh the anecdotal problem reports by a fair amount, even in spite of the well known principle that usually complaints are overrepresented on message boards and comment threads (satisfied people don't care enough to post, pissed off people do).
So unless you can provide data of your own I'm gonna call paranoia, since that link is one of the best I've found on the subject and the guys at CH are all pros, frankly I'd be more likely to listen to them than some random "Works4Me" Internet Post.
The things
I created some back-of-the-envelope predictions in July 2009 about the cost for 10TB of storage using either type of drive technology. Unfortunately, neither technology has kept pace with my predictions, but SSDs are making much better progress.
Actual July 2009 Prices for 10TB: Platter = $750, Flash = $28,125
Actual June 2012 Prices: Platter = $567, Flash = $8200
Previous Prediction for July 2010: Platter = $528, Flash = $9,868
Previous Prediction for July 2012: Platter = $262, Flash = $1,215
Previous Prediction for July 2014: Platter= $130, Flash = $150
Previous Prediction for July 2019: Platter= $23, Flash = $0.80
It's a shame to see that after three years, the prices are closer to where I hoped to see them in a single year. I think it's time to update my predictions based on what has happened over the previous 35 months. (Yes, I know this in unscientific and silly!)
New Prediction for July 2012: Platter = $562, Flash = $7916
New Prediction for July 2013: Platter = $511, Flash = $5188
New Prediction for July 2014: Platter = $464, Flash = $3400
New Prediction for July 2015: Platter = $422, Flash = $2228
New Prediction for July 2019: Platter = $287, Flash = $411
New Prediction for July 2024: Platter = $178, Flash = $50
These predictions seem much more achievable than last time. In fact, I expect that platter drives will exceed this pace as the industry recovers. I can't believe that platter drives will only see around a 50% price reduction per TB over the next seven years. However, that's been the pace of improvement from July 2009 until now.
The most interesting date will be when the technologies reach price equivalence. This would be August 2020 according to my model, at the price of $260 for 10 TB. My gut feeling is that equivalence will be reached a couple of years earlier than that, but who knows? We'll just have to watch and see!