Most reliable ssd?

[razaice]

I'm pretty ignorant with ssds. What's the most reliable ssd? I was going to buy a vertex 2, but I read an awful lot of newegg reviews saying theirs broke. Also, I only have sata 2.

 

[jkyle]

Intel by far no one else is close to them for relibility. That said I just purchased a Crucial C300 a couple days back mainly because of all the crap I have seen about OCZ I dont have a lot of faith in them.

 

[frostedflakes]

Yeah Intel is probably your best bet. Not only do they seem to have a very low rate of DOA/failure, but Intel also has a ton of resources to throw at testing and validation, which means less headaches with system incompatibility, waking from sleep issues, etc. that other controllers can suffer from. They're a bit expensive and don't always offer the highest performance, but if you want a really solid and reliable drive, you can't do better than Intel IMO.

 

[acole1]

Intel by far no one else is close to them for relibility. That said I just purchased a Crucial C300 a couple days back mainly because of all the crap I have seen about OCZ I dont have a lot of faith in them.

I just recommended the C300 to a friend after I read up on the Vertex 2 and saw all the bad reviews on NE. A chart I saw in my search showed Intel with the highest reliability.

 

[SickBeast]

What I don't understand is how the OCZ drives can be less reliable than the Intel ones when several of them use Intel memory chips.

I'm sure it's possible that the different memory chips have different lifespans, but I don't think it makes sense to necessarily bash particular SSD brands.

I suppose the controller chips could always burn out, but really it seems far more likely that the NAND chips would be the problem.

 

[mosco]

What I don't understand is how the OCZ drives can be less reliable than the Intel ones when several of them use Intel memory chips.

I'm sure it's possible that the different memory chips have different lifespans, but I don't think it makes sense to necessarily bash particular SSD brands.

I suppose the controller chips could always burn out, but really it seems far more likely that the NAND chips would be the problem.

Why does that seem more likely. SSD controllers are relatively new, there are bound to be issues. Just go check the change log on the vetex 2 firmware and the known issues. I believe there have been several issues they have fixed that they claim could have caused drives to freeze/die, etc. It's not hard to imagine that they haven't caught all of them.

 

[frostedflakes]

Yeah from what I understand most of the SandForce failures are due to something causing the controller to go into panic mode, as opposed to an actual hardware failure of the controller, flash memory, or whatever.

 

[SickBeast]

Why does that seem more likely. SSD controllers are relatively new, there are bound to be issues. Just go check the change log on the vetex 2 firmware and the known issues. I believe there have been several issues they have fixed that they claim could have caused drives to freeze/die, etc. It's not hard to imagine that they haven't caught all of them.

It's just that we know that many of the NAND chips can only be written to 3000 times. That's not very much. You don't really hear about regular HDD controllers burning out. Typically with a HDD, the platters or the heads will go bad.

It would be nice to see a comparison of the longevity of the various brands of memory, along with what happens as the manufacturing process gets smaller. It would be great to have a database.

 

[frostedflakes]

3000 per block, but keep in mind that any reasonably sized SSD is going to have tens or hundreds of thousands of blocks to spread wear over, so with good wear leveling you have a pretty large number of blocks to spread wear over. And it's less of an issue because capacities are going up at the same time, so this can offset the decrease in rated P/E cycles by giving you even more flash to spread wear over. Although another thing with 25nm that is going to hurt durability (more so with controllers that have poor write amplification) is the change from 512kB block size to 2MB.

 

[Emulex]

okay :
indilinux/sandforce = make their own controllers and firmware - do not have a hand in flash - think white box pc

intel/micron - they make their controllers/firmware/flash (intel micro joint deal) - think brand name

ocz/corsair/owc/etc can't force sandforce to give them good firmware - sandforce has moved on to better times - sandforce can't force ocz/corsair/owc/etc to use anyone particular flash - they buy parts du jour and go to town.

I know for a fact i can rely on an hp server with all hp parts working flawlessly once burnt in - i know for a fact i'll never get that level of quality (unless lucky) if i roll my own white-box. Given the pricing is similar - what would you buy?

Intel also took the safe road and did not use write-back caching to get their speeds. This was a great move on their part and the other guys in their quest to be benchmark queens - took the dangerous path.

I'm sorry - but you guys remember the seagate 7200.11 fiasco - ssd's should be more reliable. There is no excuse for sandforce and people that sell it should be ashamed

 

[SickBeast]

Wow, that's harsh.

While I agree that the Intel and Micron drives are *probably* better in terms of reliability, I don't think that the difference is as great as you make it out to be, Emulex.

Like I said, if the NAND chips are identical, there's not going to be a significant difference in the drive reliability, IMO.

I had a choice between a 40gb Intel x25-m, a Kingston V100 64gb, or an OCZ Vertex 2 60gb, all for around $85. I could have gone with the Intel, but I would have sacrificed a lot of capacity in doing so, and I don't think the performance would have been too different either way.

 

[Emulex]

they aren't identical. Micron makes the best nand in large quantity. intel has majority stake in the deal. so between micron and intel - there is nothing but scraps left for the rest. demand far exceeds production esp with enterprise slc/mlc - everyone wants - at any cost.

samsung is tied up with apple $$$. plenty of scraps for the rest of the guys to piece together with various firmware with various controllers.

You will never get fired for buying intel

 

[groberts101]

I guess if the Sandforce's OS performance potential wasn't so good(compatibility issues aside).. and the controller wasn't also a main factor in the mix?.. I would have to agree.

But with as many Sandforce controllers out on the market now I can't say that all wish they'd gone Intel. I ofetn see many compare the 2 and are happy to be on Sandforce.

Seems like the best advice comes from those that have used both, IMHO.

 

[Dark Shroud]

I had a choice between a 40gb Intel x25-m, a Kingston V100 64gb, or an OCZ Vertex 2 60gb, all for around $85. I could have gone with the Intel, but I would have sacrificed a lot of capacity in doing so, and I don't think the performance would have been too different either way.

The difference is that Intel has a better quality control process as well as making both the controller & firmware. So for the current range of SATA 3gig drives the Intel 320 series are the over all best.

Intel also is the only company to rate their 25nm NAND for 5,000 instead of 3,000.

OCZ now owns indilinux so they can make their own controllers & firmware for the controllers. But they still have to depend on Samsung for NAND.

The other companies can still use Marvel for their controllers. I don't remember if Samsung makes controllers.

 

[Emulex]

exactly you buy a intel cpu, intel chipset mobo, intel ssd = good to go. guaranteed.

Intel makes it all. now with storage they will dominate the world(servers). intel xeon's/intel server chipset, intel enterprise mlc. buy their stock!!

apple gave up on rack servers - they want folks to rock mac mini's lol - pita to rack up no hot swap power supplies - the server war was has won. intel is the emperor and lord microsoft is at his side.