I think you need to read the article in detail again. The figure is 10,000 erasures of any one block before that one block goes bad. And that's an estimate, lowballed for lawsuit resistance.
Erasures are far less common than writes, or even random writes. The drive's electronics and "load-leveling" take care of that. Yes, eventually you will run up against the limit, and blocks will start going bad, little by little. That, however, is just like a good drive equipped with SMART. You'll start to lose capacity and eventually the BIOS will tell you "Wait a second! Time to replace the drive."
MTBF on these drives is rated in decades. The article squawks about the performance decrease, which can be mitigated (especially effectively in the Intel drives) by doing a low-level erase using a special tool outlined in the article. Once you "season" the drive in that way, you reset the location-leveling logic and recovery most of the performance of the drive.
Also, I'm not talking about TCO because TCO doesn't take "silly-stupid-fast" into account. If you want near-RAMdisk performance, get the SSD. If that isn't that important, don't get it. Money is always a factor that will influence that decision, of course. |
