SSD 5 years lifespan - quote from Winslow in 2009
Recently, I've been noticing issues with my Intel SSD and started to Google some very interesting articles on potential failure rates, including this article. I find it rather disheartening that they only last 5 years for such a high cost but also hearing failure rates of up to 20%.
Intel boosts speed, cuts prices of solid-state drives | Nanotech - The Circuits Blog - CNET News
Incoming AKG's reply..... lol
I have not been impressed with the reliability of my SSDs. I have 2 vertex, 1 agility, 1 intel Gen2 and one no-name IDE SSD. I have had to RMA two of these and I have had data corruption issues on 3 of them. I have also had imcompatibility issues as well. The speed is great but I wouldn't trust them with critical data at all. The only SSD I have had no troubles with is the 30GB Vertex in my HTPC and I have feeling that is because the only thing it does is boot to windows a couple times a week and that is it (all media is streamed to the HTPC).
I think the issue is writes. Having to write to a cell requires higher voltages to "impregnate" a bit into that cell, which can lead to degradation. If you use your SSD as only an OS drive then you shouldn't be having too many issues. If you download a lot, save your files to hard disk directly, instead of pre-storing them on the SSd and then transferring them over. Also, now is a good time to look at TRIM, as it does save a lot of hardcore writes over time.
That is the funny thing, my SSDs are mainly used as either an OS drive (with My documents, download, etc moved to a HDD) or as my games drive (with saved games folders moved to HDD as well). I also use trim on all my drives as well.
Guys....do you really think you write 25GB of data to your SSD in a day? That is server level disk access. IMHO Intel is trying to push MLC into the server environment....which stupid. BUT that is what Intel's main market push is right now and thus that quote. They really dont care about consumers or at least dont spend much of their marketing budget and time on us. IMHO: MLC == SOHO and consumer SLC == Server no matter what a spin doctor says. ;)
The biggest thing that will kill your SSD is filling it to capacity. You really need to leave 10-20% or so free so wear leveling can do its magic (TBH no matter what size you are running leave a min of 10-15GB free). MORE free space is better. Unlike Hard drives which get slower if you fill them up (and access that data), SSDs dont get slower BUT they trade lifespan for speed when you fill them up. IF you are using the 30gb'ers. Drop the OS on it and maybe a program or two (consider them really 20GB of usefull space). THAT is all they are meant for. The 60s are more forgiving as you got 45gb of room but 50 is max I would feel comfortable with. For 80s I consider 65 about the most I would use. The 128s....I recommend a max of 100GB. Like I said exect (on avg) the bigger ones to last longer then the smaller ones.
IF you have numerous cells dieing in your first year...you over filled the drive or got a bad one. As with all mass produced items the average may be 10k....but some are going to die faster then others. It is better to step up a size and down a step in performance then get the fastest small one you can afford. More free cells is better! Going hand and hand with this is you need to either run TRIM or manual TRIM / wiper on your SSDS. TRIM is less harder on your cells then Idle Time Garbage Collection which chews through cells at a copious rate. ITGC really does burn more cells (I run ITGC as it is faster....but do so understanding it will kill the drives quicker) and if you let it get into a semi-degrade state the controller WILL kick in ITGC to clean up the mess. About the only exception to this rule is Intel, whose ITGC is so mild as to be all but useless (BUT is easy on the cells).
The other thing that will kill your SSD is dirty power. I really do have to laugh (its either that or cry) when I see peeps running on noname PSU's yet have a kick arse SSD in their system and wonder why it died so fast. Its not just no-names there are plenty of well known PSU's that I wouldn't touch with a 10ft pole that have a good rep. Low ripple = happy SSD. This is not just a SSD problem as ripple kills HDDs to...just slower (in other words they tell you when they are unhappy much quicker then a HDD would....which usually takes it silently for longer before going POOF!)
If you are running your ssd on an older OS and didnt do a diskpar / diskpart to align the cells you will kill it faster then otherwise. Aligned SSD = happy cells. NON Aligned SSD = higher cell usage....which will kill it faster.
All this aint aimed at you Varroa. Just general comments on proper care and feeding of a SSD.
Personally, I don't think we can "assume" the life span of SSDs until they have been on the market for at least five years.
The one thing I do want to point out is that several VERY early adopters got screwed. Many of the people that bought the first netbooks that were released with SSDs are experiencing serious performance degredation and nothing they can do will fix it as this point unfortunately.
That issue is now moot though, as first gen SSDs are all EOL for awhile. Modern SSD (gen 2+) have ways of keeping them fast internally. TRIM and ITGC are what keep them fast and keep this from happening. Basically with these older drives, if your SSD is slowing down it is because it is in a degraded state and has no free cells. The fix is easy (yet does cost cells). Run FreeSpaceCleaner w/ FF option (or several other programs that basically do the same thing) and get the cells back into a virgin state. It is a work around and is a kludge....but that is the cost of being a super early adopter.
Akg or anyone else familiar; can you either explain or direct me to a site that explains why programs such as
FSC is what I've been using in conjunction with Perfect Disk 10 for my Intel SSD 80 gig raid-0 ver. 2 with the latest firmware (2CV102HD) (Write-back cache disabled ( default)).
I've since installed Intel Rapid Storage.
They dont damage the drives per say. What they do is write FF to all free cells (whether they are degraded or not). This obviously uses up a cell life in doing so across many more cells then a TRIM does. It is more harsh and less precise then TRIM, because TRIM just cleans the cells that NEED it (by basically writing FF to all the degraded and no longer used cells). What you are doing is basically a big hammer version of ITGC...and is even harder on the cells then the typical controller level ITGC. Basically while it is called "Tony Trim" its more like Tony ITGC. As the first part consildates data (thus using writes) and then the FSC /w FF option does even more writes.
In a nutshell you are writing and then rewriting to cells every time you do that 1-2 punch and using up more cells in the process. It is all you can do where you run raid....but this is one of the reasons I dont like raid ing SSDs and hardly recommend it to anyone.
Basically on avg for all ssds from less harsh to most (with trim being the best, lowest cell usage solution):
TRIM > ITGC > FreeSpace cleaner > Tony TRIM. (though the dif between T Trim and FSC is small, neither are really a good idea on modern SSDs unless you have no other choice).
|All times are GMT -7. The time now is 08:49 PM.|