SSD (obviously) works under a different premise than a mechanical drive. Is it harmful to the physical aspects of a SSD to have constant 2 to 10gb updates? It seems, from my understanding of the mechanism, this would be damaging with constant overwrites and re-appropriated data. Would this cause bad sectors? Would it "burn holes"? Am I totally off-base and misunderstanding? Please explain why or why not.
Comments
I don't think there is any problem using a modern SSD a lot. I have windows and most games loaded on an SSD on my system.
------------
2025: 48 years on the Net.
Constantly writing stuff do eventually wear out any disc, SSD, SATA or SCSI but we are talking about many years here. SSDs are less heat sensitive and you don't get that issue where a tiny bit falls off something and slowly destroying sectors from you.
There have been problems with some SSDs a few years back (they tended to just die) but it had nothing to do with this thing, it was a hardware issue.
So this is either a myth or something a few really early discs had.
Harddrives do die sometimes, I think I killed 12 or so regular discs and one SSD (I had maybe 60 regular discs, 1 SCSI and 8 SSDs. It is always a risk to not backup your stuff or use a RAID partition (I prefer RAID 5 with 4 discs, fast and safe) but SSDs are not a greter risk then any other type of drive.
A few thousand writes across the entire drive is a lot, though. Whether "constant" 10 GB updates are a problem depends on how frequent you mean by that. Writing 10 GB per day to an SSD isn't likely to cause trouble, as it will die of something else before a write endurance failure. 10 GB per minute will wear things out pretty fast, though.
Writes do kind of "burn holes" at an atomic level (by forcing electrons into or out of an area that is designed to stop electrons from randomly leaking through), but again, you get a few thousand writes to a page before the damage adds up enough to be a problem.
SSDs employ sophisticated wear leveling algorithms to try to spread out the writes. So even if you write to one file a million times, that's not going to hit the same page a million times. The SSD will physically move the file around so that you're hitting a bunch of different places few times each.
SSD do not understand what files are flagged as deleted by the OS and which are not. (when you delete a file in windows, it is not physically deteled, it is flagged). Each time you write to an SSD, you are actually doing an overwrite, unless your SSD is new.
So the more data you use on your SSD, the more writes turn into overwrites. If you ever fully fill your SSD, every write operation = overwrite. That's why they slow down after a while.
But since Windows 7, the OS can send TRIM commands to the SSD, when the SSD is idle, garbage collection is handled during idle time now. So when you're doing nothing on your PC, Windows is telling the SSD which flagged files are actually deleted files.
Writing to an SSD will always degrade it, just far less since Windows 7. SSD have a limited amount of write cycles, far lower than HDD, TRIM just makes the drive slow down less.
"We all do the best we can based on life experience, point of view, and our ability to believe in ourselves." - Naropa "We don't see things as they are, we see them as we are." SR Covey
SSDs have a way to measure how many times each cell has been written on. Once a threshold is reached, that cell won't be used any more. SSDs also have some extra cells reserved for degradation, so that when one cell is used too many times a reserve cell gets used instead, and usable drive size won't get smaller.
You won't burn a hole in SSD even if you constantly keep rewriting same data. You'd have to use SSD so much that the whole SSD would wear out to break it.
Techreport.com did an endurance experiement with six 240GB SSDs. They were able to write more than 600 TB on all of them (that's 600 000 GB). It's possible to wear an SSD out, but it's not easy.
http://techreport.com/review/26523/the-ssd-endurance-experiment-casualties-on-the-way-to-a-petabyte
SSDs have an advantge in that there are no moving parts. You can drop an SSD and nothing will break (I'm sure if you drop it off the Empire State Building you may crack the PCB, but it's not like a HDD where you could just bump itthe wrong way and wreck the drive). The limited number of writes is correct, but all SSD firmware have this in mind, and use a myriad of functions to extend life. Overprovisioning, compression, and wear leveling are the three most common. Because of physics, SSDs are much more sensitive to higher temperatures - they will degrade faster at extreme temperatures. Now, most SSDs are low power devices, and aren't really going to get very hot in a typical computer case, but if you stack 30 of them in a server chassis with limited air flow, you could experience problems.
SSDs haven't really been around long enough to get really good reliability data, but the "common sense" notion right now is that SSDs are not really any better or worse than traditional HDDs with regard to failure under tpical consumer-level loads. They seem to be on par last about the same length of time - which is an average of about 5-6 years, with some failing well before that, and some (presumably, we haven't got there yet) lasting a lot longer than that.
The undisputed facts are that SSDs are a whole lot faster for typical consumer loads, and that they are more expensive per byte.
Modern MLC SSDs are rated for 2000 writes per cell. Assuming you have a 250 GB drive and write 50 GB per day, it will take you 5 days to write to each cell once, theoretically. At that rate, the drive should last 10,000 days, or just over 27 years. Do you really write 50GB to your SSD every single day? That's a brand new game, and a fairly big one at that, every day.
If that doesn't convince you, here is some destructive testing performed on SSDs. Note the Samsung 840 uses TLC memory, not MLC, and newer TLC memory may have better write endurance.
http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead