SSD - die after 1 year?

[lolage]

Hi

One of my mates has got a SSD but apprently they only last about a year - is this correct or is it down to the brand or what? Seems stupid spending all this money on SSD for it to break after about a year??

 

[rpstewart]

While it's true that the memory cells in SSDs are only good for a certain number of write cycles they will last for far more than a year in normal use.

For example a 100Gb drive with cells that will do 100K write cycles (which is a reasonable estimate of the life of current devices) could survive having 10 petabytes of data written to it before having problems.

 

[VaderDSL]

It depends on write cycles I think? NAND flash has a limited number of times you can write to a cell in it.

If you're a general user, just using it 24/7 365 days a year I've heard it'll last much longer.

Let's quickly do the math again. If you have a 100GB drive and you write 7GB per day you'll program every MLC NAND cell in the drive in just over 14 days—that's one cycle out of three thousand. Outside of SandForce controllers, most SSD controllers will have a write amplification factor greater than 1 in any workload. If we assume a constant write amplification of 20x (and perfect wear leveling) we're still talking about a useful NAND lifespan of almost 6 years. In practice, write amplification for desktop workloads is significantly lower than that.

 

[PhillyDee]

That used to be the thought, but just isn't true. Your 'mate' is just trying to show off.

 

[Vore]

People over-react to the apparent 'short' lifespan of SSD's. Granted, with unusual usage patterns it could potentially be an issue (hundreds of program/erase cycles per day) - but for your regular joe out there they will last for a very long time.

Here is an exerpt regarding OCZ Vertex 3 drives and MTBF for reference:

"The SSD NAND FLASH partitions

So the storage unit we'll be testing today is a 240GB version. Here's how that works. OCZ places a NAND flash partition of 16 ICs onto one PCB. The NAND FLASH partition the new SandForce 2281 assigned, opposed to the 2582 Controller used in the PRO assigned for multi-channel IO.

Main controller differences, the regular Vertex does not have some enterprise features like power loss data protection & data recovery protection, it also offers 28% over-provisioning for optimal drive endurance and data protection & management hence the PRO version has less available storage space. Now there's a 20% IOPS perf increase over the last generation product for the Vertex 3 but 30% for the PRO.

No, the biggest difference in-between the regular Vertex 3 and the Vertex 3 PRO is the fact that the PRO makes use of Enterprise-Class MLC NAND Flash, yes .. better NAND MCL Flash memory. As such it got a MTBF of 10 million hours (!) and the regular Vertex 3 will get the new (cost effective) 25nm NAND Flash memory with a lesser lifespan set at an MTBF of 2 Million hours.

So what about 25nm NAND lifespan ?

Personally I'm not at all a excited about the new 25nm NAND FLASH memories, the overall lifespan of the ICs has been reduced from 10.000 towards 5,000 program/erase cycles. Rumors are that the numbers for consumer grade 25nm NAND flash memory (as used on the SSD tested today) are even lower at 3000 program/erase cycles.

But granted, as drastic as that sounds, it's all relative as this lifespan will very likely last longer than any mechanical HDD. Drive wearing protection will help you out greatly. With a normal filled SSD and very heavy writing/usage of say 10GB data each day 365 days a year you'd be looking at roughly 22 full SSD write cycles per year, out of the 3000 (worst case scenario) available. However, all calculations on this matter debatable as usage differs and even things like how much free space you leave on your SSD can effect the drive."