mkozlows: Proably better not to try and find any data in the anecdotes.
Warren: what? SSDs don’t fail because the cells die of old age. The entire point of this thread is that they’re dying far earlier because of controller issues. That’s what “doesn’t show up in BIOS” means. You can write all you like, as Anand showed in his articles waay back in 2009 other components will fail before you hit the wear limit.
Yes, but with AppleCare, it’s paid for by Apple after 1 year (well, 2 with my credit card extension).
The new USB 3.0 drives. How fast are those compared to a normal 5600 rpm SATA II drive? Do they get as much wear as SSD’s? Maybe the best option is to tun USB 3.0 drives to handle commonly thrashed items like your temporary internet files, swap file, e-mail, all temp folders, and restore points.
Any thoughts or graphs?
My thought it is that commonly thrashed items and USB drives don’t seem like a natural fit, but I’m not sure what you want from it. What would you gain by putting the items you mention on an SSD? The quality of the wear leveling varies too much to say something about it in general, although I’d expect it to be worse than comparable SSDs.
I don’t have any graphs though.
Honestly your best option is to have at least 6-8GB of ram, offload your temp and swap file to an HDD and not worry about it. The performance impact you’d get from your swap file being on the HDD is minimal, since it’ll rarely access it.
What is a good backup strategy for a home user? I got a Terrabyte drive, about 40% full, and its a normal HDD not a SSD. Its not like even burning DVDs would be a good way to backup my drive. The only thing I can think of is a mirror volume although I would like some other way.
I was considering getting an SSD at some point, but not after these articles. Ill wait quite a few more years until the bugs are all worked out.
I was just curious if USB devices hit a speed and latency rate that equals mechanical drives. The request for a graph was if someone had done for comparison tests (which I haven’t found anywhere myself).
I had a SSD in my home desktop for about 3 months and it’s still going strong. I use it for OS and games.
I love it and is considering a SSD for my work laptop. I guess it is not a good idea. I wouldn’t dare lose the work data!
Weird. I expect a work laptop to have a proper working backup system in place, so it’s the safest computer to use an SSD on.
Nope. We had to buy our own DVD for backup. How’s that for state of the art?
Oh boy. In your case, I’d expense an additional USB portable harddisk to be the “live” backup drive, or even the main data drive.
The only hard data I’ve seen indicates that at least Intel SSDs are more reliable than consumer hard drives, so I don’t see why you’d be particularly concerned about them at least. You might want to avoid non-Intel SSDs, and if you are worried about losing data I don’t see why you wouldn’t be using backups already.
If you have an Intel SSD, download and run this:
It’ll tell you if you’ve got something configured stupidly on your system.
I came across the statistics I mentioned in the other thread again. These numbers were from items sold April-October 2010 and RMA:d before April 2011, with at least 500 sold. It’s in Swedish, but all on one page: http://www.sweclockers.com/nyhet/13859-fransk-datorsajt-publicerar-rma-statistik-for-april-manad unlike the French version which is linked at the end. The short version:
SSD
Intel: 0.3%
Corsair: 2.7%
OCZ: 3.5%
HDD
WD: 1.5%
Samsung: 1.8%
Seagate: 2%
Hitachi: 3.1%
Pretty good numbers for Intel there. I suppose that’s why they’re upping their warranty to five years.
Part of those OCZ numbers would be the debacle with the Vertex 2’s and the shift in NAND causing performance drop and size decrease. They were doing free swaps for anyone who wanted the older NAND for better performance/size.
That being said, the numbers are pretty reasonable and hardly something to be freaked out over. Really not that much worse than an HDD, and in Intel’s case, a lot better.
I see; that might explain the difference between the numbers from OCZ themselves (2% I think) and the RMA numbers, although it is also possible that the period was different.