News 1 in 3 SSD failure rate 'unfounded'

http://www.bit-tech.net/news/2008/03/20/1_in_3_ssd_failure_rate_unfounded/1

Avian Security published a report last week claiming that solid-state drives were experiencing a 30% return rate, but box builder Dell has called that "wholly inaccurate."

 

Can't for the life of me remember where i saw the link now, but i remember recently seeing a news piece about the SSD return rates:

Yes, they were getting a hefty return rate, but not due to failures, instead due to not performing as well as customers expected for the considerable extra price.

which would make a LOT more sense

 

Yeah, I seem to remember seeing some news exactly like that regarding the macbook air. For 1000$ extra i'm not surprised.

 

Isn't it the random access speed which really kills SSDs? I would imagine having a pair in RAID would be killer though

 

No - SSD technology has a controller chip inside that varies where the bits are written to minimise wear. They don't crash like normal hard drives unless there's a catastrophic failure (unlikely - no moving parts) they just gradually lose capacity after a certain amount of time.

 

Bindibadgi i think it is minimize wear and maximize life time....

 

Nope. Random Access speed is SSD's party piece (It has a 0.00001ns random access time). At the moment, it's bandwidth. While a hard-drive may take 200ns to get from block to block, it can then transfer that block at a full 150Mb/s. SSD's on the other hand, get to a block instantly, but can only transfer at 33Mb/s (Or more, or less depending on the SSD).

 

LOL yes, that's what I meant

Ive seen SSDs claim 100+MB/s now - OCZ's for example comes close iirc

 

whats the bit-tech staff experience of failure with SSDs

 

I should be able to give first-hand experience of raid0 with two SSDs soon.
Friend of mine ordered two of the 32GB Sandisk ones from overclockers.
Unfortunately they're out of stock at the moment with no ETA for more coming in.
Still, watch this space.

As for high failure rates, I can't imagine how that would even be possible.
There's no moving parts so even dropping one while its on would be no worse than dropping your solid-state MP3 player.

 

I work for Puget Systems, a botique integrator. Out of a dozen or so SSDs we've had problems with two, I think. One definitely doesn't count, as it was a very early release sample with a bug that needed working out, and the other sort of doesn't count - we've been having issues getting benchmarks to run smoothly on it, but that's a problem that we're catching and taking care of long before it gets to the customer. We haven't had any fail after delivery to the customer. Overall, I wouldn't worry about the reports of high failure rates. There's always some weirdness that needs to get sorted out at the start of a new technology's lifecycle. Between that and the price, I generally think it's worth being six months behind the bleeding edge.

And yes, we've sold SSDs in RAID. They can do some pretty awesome stuff, but it costs as much as the rest of the computer!

 

i don't think that is true as well.

we have been hearing FUD here and there. it feels like there's something going on in corporate back rooms or some misunderstanding.

initially, some executives in HDD recovery businesses were quietly telling us SSD crash will mean data is irrecoverable. that's full of "bollox" - borrowing a word from someone i know.

believe it or not, SSD has hit SATA-1 speed and beyond, as of now. We're not seeing them in market, yet. Some smarter SSD controllers are there already.

Prices are dropping in Japan market for certain type of SSD that uses less reliable MLC NAND (as oppose to SLC NAND) but they put a lot more into the package to makeup for the faster failure rate. These "backup" are hidden. At the end, it is expected to work for 5 years. I think Toshiba is doing that and they got nice controllers.

Rubbing my hands in anticipation. at least 6 more months baby, 6 months.... :O

 

The early sample Samsung one we reviewed last year was extremely slow and designed for notebooks (we were just wondering if it made any difference to PC use with zero random read cost) but that still works just fine. When prices come down we'll definitely look at some more

TR look at SSDs in notebooks now and again and they've never had a problem afaik.

 

yep the relatively cheap ones (like the £200 ish samsungs) are pretty slow and only really worth having for laptops or if you have a specific need imo
the really expensive ones like the MTrons and the OCZ ones are fast but holy crap are they pricey

reckon another year or 2 and they'll be a worthwhile buy

 

I work at a software house, and my managing director is bonkers about having the fastest laptop.
He got a 64GB SSD drive, and he was very very happy

until it broke down just before a demo and had to swap back to a normal harddrive

He got it replaced, then it broke again...

Make your own judgement..

 

By that reasoning all hard disks are unreliable the death stars (deskstar) didn't get there name for nothing.

 

there are a few laptop hdd -> Compact Flash adapters out there. I'd just use one of those.

 

Bad batch, crap brand, unlucky?

 

I'd say it's probably true.

The consensus seems to be that if you keep Windows power settings on the default Sleep mode, instead of changing to Shut down, your SSD will fail withing a few months. At least if you have a Sandforce controller. My OCZ lasted two months in a new Dell Vostro 3700 laptop with the latest A10 BIOS before starting to turn out BSODs, and three months before failing completely.

This conclusion can be reinforced by the fact that the OCZ Forum has an official recommendation to change Sleep to Shut down in Windows.

Since Sleep is default most people will use that, and if Sleep cause failures I wouldn't be surprised to see 30% or 40% or 100% failure rates.