Is SSD really worth the price or is it just a rip-off?

I'd be worried about one of the platters coming off the spindle and decapitating me at that rotational speed

 

That the price I'm looking for. I'd imagine it's a function of how quickly the chip manufacturers can ramp up production.

 

Right, and my thought is that I'm likely to use this drive on more than one laptop in its lifetime.

 

I agree. I have actually had a close encounter with rouge rotating storage media.

Many years ago I was sitting in front of my desktop PC (nowadays I only go laptop).My CD-ROM was having problem reading a CD-R and I could hear how the CD-ROM drive would spin the disc down and then up again. After waiting a minute or so, listening to the CD-ROM drive going WHIIIRRRR *pause* WHIRRRR *pause* WHIRRRR *pause* etc., I gave up and realized that the CD-R was damaged and I decided to give up on it.

As I am pressing the eject button of the CD-ROM drive, I hear the CD-ROM going WHIRRRR, starting to spin-up the CD-R again in its attempt to read the disc. My mind manages to think “Uhh .. ohh” as the CD-ROM drive opens. Out of the CD-ROM drive comes the damaged CD-R, wildly rotating and flying right towards my face. As the distance between the CD-ROM and my face was short (maybe half a metre) I have no time to duck. The rotating CD-R slams violently into my forehead and then bounces away, flying a good 6 – 7 meters into another room through an open door.

I do not start to bleed from my forehead but I do get a nasty red blotched line, where the disc struck me. Later that day I meet up with my mates, who ask me what is up with my forehead. I tell them the story and they die laughing. From that day they started calling me Mr Disc Head.

 

To get back on topic...

I blatantly ripped the table below from a Sandisk manual on their PDFs. The table shows MTBF data for the different Sandisk SSDs. As you can see, the MTBF is drastically lower for higher capacity SSDs than for low capacity SSDs.

Of particular interest is the data for the 128GB version, where the MTBF is specified as less than 500 000 hours, which actually seems to be on the low side.

(I do not understand why the table is shown so far down in this post but I cannot do it any other way).

Capacity	MTBF (Hours)	FIT (Failure Rate per Million Hours)
1GB – 8GB, 16GB	1,485,397	0.6808
32GB	1,056,635	1.4037
48GB	871,839	1.1470
64GB	547,105	1.8278
80GB	601,214	1.6633
128GB	471,209	2.1222

 

I wonder what kind of testing methodoligy they're using to get those numbers... point is, SSD's have a high failure rate, no matter how the numbers get twisted. I'm just still not convinced that the technology is ready to replace HDD's. Flash memory has been around a very long time, but it was never meant to be used in such extreme environments with constant reading/writing (i.e. in a hard drive), and until they (the manufacturers) get over that hurdle, I would never consider buying one.
Here's one story, where a guy used several of them in a server environment, and within 10 weeks 30% of his SSD drives failed.

 

talin:

Just as with any technology there is the probability of a manufacturing defect for a particular lot causing excessively high failures. Given the volume (ok, really lack thereof) I would say that it is quite possible that these could all be from one manufacturing run.

Given the low capacity available at the time (32 GB / drive) I can't see the price/performance being worth it for a server.

There are a few other things I find suspect with the article, including the poster's misconception of MTBF.

Cheers,

 

I'm still not convinced John. It's the way of the future sure, but for now, I think there's room for a lot of improvement, namely reliability, and price.

 

Flash memory was not meant to replace HDDs for a simple reason : cost. Back in the time where you paid a hefty $100 for your tiny 1GB USB key, would you have ever thought about : "Wow, for $2000 I could afford a slow 16GB drive" ? Certainly not, but the idea of SSDs was still around, with people trying to assemble CompactFlash cards...

Then the prices fell down, and it became economically viable to develop this kind of memory. That's all.

Yes, there is still room for improvement, especially in price. But as for now, we have drive that are extremely performant, even for power-users. I don't think that reliability would be a problem (maybe with thinner engravings and all those stories of neutrons ).

 

dude its a rip off.. i remember 5 years ago seeing a online that sold a thing that would let u plug ram sticks into a set up and have it work as a 10-40 gig hdd... and it was a fraction of the price of the SSD u see now i remember a 10gig pre installed thing was 700 tops ... it prob was not as fast but still..... remember how high ram prices used 2 be?

also how much does it really help battery.... i have not seen any one bragging about how much it helps...


Imo use a USB pen... lewl

 

Anecdotes aren't actual failure statistics. Even if the story is true (no way to tell) that says nothing about the aggregate failure of the technology as a statistical group. Dell has denied claims that SSD based systems have a higher failure rate than systems with mechanical drives.

I agree with you on one side of the equation, price. However (like CPUs) they're subject to Moore's Law so the technology is just going to get cheaper, with higher density and likely faster too.

This is the early side of the adoption curve, with high prices and the other associated "fun". It will be quite interesting when these reach commodity status.

Cheers,

 

I think you need a memory pricing refresher. Here's a site
that tracks historic RAM prices. 5 years ago, you couldn't buy a 1GB stick and the pricing for a 512MB stick ranged from $61 to > $200 depending on speed, capacity and ECC vs. non-ecc. To hit 10GB, you needed 20 sticks to hit 10GB, and that means > $1200 for the raw memory alone assuming you were buying cheap non-ECC memory. 40GB, sweet. That would be in the ballpark of $5000 before you add the interface device.

Battery life is a nominal gain (if any). More tangible benefits include significantly faster access times, lower heat production, lower weight and greatershock resistance. I don't own one yet as they still aren't quite cheap enough to spend my personal $$$ on them, but I suspect it will be by year's end.

 

John I fear you may have started a rush of novices trying to buy ECC RAM for their notebooks because who wants the cheap stuff? I hope not.

 

I really cannot understand how anyone can question reliability. This is not a new science. Its been used for years in business and enterprise environments and they don't question reliability whatsoever. I have been watching very closely for valid reports of ssd malfunction without any luck.

If someone is going to question reliability, for my sake, please back it up with some facts so I can understand why its even in question. It a learning process for me as well.

 

May be newer versions will be cheaper and more worth it.But as of now,i would say no.

 

Well Les, it was backed up especially with all those reports that Dell of course denied. But who am I to question a company's wisdom? Afterall, they're just out to make money, right?

 

Do you really want to do mechanical disk replacement without ECC? I wouldn't. Not that non-ECC is terrible; but any data corruption is bad

Cheers,

 

That's a personal value proposition. I quite respect that and we all have tipping points to purchase based on price. It's still a little more than I'm willing to pay at this point in time. In a few months, who knows? Competition, manufacturing volume and Moore's law will work in our favor over time.

My issue has been with the "ripoff" statements, and unsubstantiated failure claims. I don't doubt that there are failures; it's technology after all and it isn't infallible. The magnitude of claimed failure is where I'm having issues.

 

Not right now. Once the prices go way down then yes.

 

Are there really "all those reports" or is it one report that's cited over and over?
All too often in this internet era there's only one report that gets cited time and time again. Each citation of the same report gets counted even though it's really only one report.

 

It could be, John. I just air on the side of caution. I'm not against new tech by any means, but I couldn't recommend to anyone that they be beta testers for companies, especially at these prices.
Flash memory was never meant to be used in such a way, that the memory was written to over and over and over frequently, and until that's overcome and made reliable, I cannot recommend SSD drives to anyone.
Some are gung-ho on this tech, but I'm more cautious, that's all. Given 5 years from now, it will be established by then if it's truly reliable or not, but how can anyone say that it's reliable when it's still so new?
Give it some time and we'll see, and if I'm wrong, then great. But there are some here that are definitely biased toward it and any claims to the contrary of what they say are shot down as false. This argument will never end, so I end it here. We'll all see in a few years who was right. Atleast I can admit when I'm wrong, and if I am, I'll be the first to admit it.

 

John honestly I am confused in this one area (the post I grabbed quote from before). But the volatile memory replacing a mechanical disk causes me issues even before I get to ECC or Non-ECC. Sure I can see the viability in certain specialty implementations where of course the volatility issues/problems are addressed. But not in any Home/Office implementation as any electrical issue could cause data loss. I would guess if people already hate HDD failure volatile would make HDD look like the standard of stability.

I thought you were commenting on a previous post and running numbers for discussion. The idea of RAM as HDD replacement seemed well "stupid" to me from the onset. Please inform me as I could really be missing something here?

And for the novices out there. Any normal power user not doing specialty tasks or applications does not want ECC Fully Buffered RAM, both expense and speed imho. (1 error per GB per month. Wikki.)

 

Wait a minute here...maybe we want to check the internet regarding that. The editor of that has retracted and apologized PROFUSELY to Dell. He concedes quite clearly that Dell's TRUE accounting of about 2% (not 20%) return rate is correct and that there calculation erred in looking at an important factor which, I believe was the true total number of ssds sold.

I am more than happy to address any of the reports. If you look hard, you will also find a 'tester' who through a number of 'low quality' ssds into his network and tied to max out the wear levelling claims...

Anyone who truly doesn't believe ssds will be revolutionizing the storage industry should do a bit of history to see where they started and where they are now.

We are not dealing with a new technology. We are dealing with a proven and superior technology which can validate claims that an ssd CAN last in excess of a hundred years.

Truly, much of the dissention seems to be between those who have an ssd and those who haven't. I would surely love to hear negative viewpoints from owners of ssds (if there are other than capacity and price concerns).

Here! and here....(non-Dell response)

 

No argument from me -- the question is how detailed do you want to get Specialty apps would include OLTP databases for applications such as CC processing for those that don't know. Yes, I do think about this stuff on a daily basis.

It is unless you work in the spaces that might require that type of solution. I sometimes have a hard time differentiating between high availability and the real world as HA is real world to me.

THat's one too many uncorrected errors in some environments

 

So half of the prima fascia argument is inaccurate? No way

It's fairly easy to design a pathological case for this. Real time video capture in an infinite loop for something like casino monitoring. There you go!

In optimal conditions it should outlast the computer it's installed in. Often with mechanical drives that's not the case.

I don't have any SSDs yet. But I do understand their value and that we're nearing the end of the Winchester drive technology. Although, I could see 2 drives one online SSD and one offline backup.

Cheers,

 

What about SSDs allegedly being slower than HDs when using Outlook or video streaming?

True or false, the "rumors" raise doubts. I don't want to put that amount of money at risk.
I'll wait till it's ripe and offers enough capacity.
Cheers

 

I'd say prove it. If the allegation is made, it should be capable of being validated. Whether it's an SSD or a traditional mechanical drive they will both struggle with the series of short reads that is encountered when working with an Outlook file.

However, if the OS were competently designed the writes would be cached until exit from the program or until a page cleanup runs.

As for video streaming. That should be something the SSD excels at.

We wouldn't want facts to get in the way of a decision.

 

This Outlook problem, recognized, only occurs with SSDs which cannont deal with many I/O, such as Sandisks. And it necessitates a lot (i mean A LOT) of emails for you to notice it.

And when it would appear, your laptop will not tumble around the table, it would just freeze for a second...

 

I merely reacted on the steep decline in MTBF when moving from small capacity SSDs to large capacity SSDs. Sandisk’s 16 GB SSD had a MTBF of 1 485 397 hours, while Sandisk’s 128 GB SSD had a MTBF of 471 209 hours.

I found the almost 70% drop in MTBF between 16 GB and 128 GB to be remarkable. Also, in absolute terms I believe that today’s desktop hard drives have MTBF in excess of 1 000 000 hours.

I have no idea why there is such a sharp drop in MTBF when capacity increased in Sandisk’s SSDs. Perhaps the higher capacity drives utilize many more memory chips than the low capacity drives and the increase in chips results in decreased reliability.

 

There is a common thread in what your writing to past problems with the same company...Sandisk. I don't know why there MTBF drops so significantly but I wonder if its a mlc ssd that tyhey are talking about. They have endured alot of criticism with no response on their initially 5000 series releases as so many had pausing concerns that were hard to take.