There were 2 references to the characteristics of NAND memories as used in SSDs as I watched the online videos of the new SSD drive with a 500Mb/sec transfer. The first reference was to the need to run a program (recommended daily) to keep the access speed up, on the SSD drive. Regarding this point, is this the equivalent of a defrag program for a mechanical drive, where there are holes as files are read and written, and this puts them back together again and maybe even optimizes the written space and the blank space?
Secondly, there is the mention of backing up the drive because of the limitation of how many times a NAND can be written-to or read, before it starts going bad. What exactly is this limitation? How does it occur? Is this the identical to the failure rate of any solid state DIMM memory carded into the computer memory slots, based on mean-time-to-failure, or is something else going on here?
A third question, not covered in the videos is,"Is this non-volatile memory, or is the memory held alive by an onboard supercap or other voltage source?
There is no program that is needed to run daily to maintain SSD performance. Newer operating system support a command called TRIM which tells the SSD when to clean up deleted pages.
SSDs are not HDDS.... SSD intentially fragment data across multiple NAND chips for performance and wear leveling.
NAND chips eventually wear out after 3000-5000 writes. The layers in the cell eventually wear out. However, this should not be a big worry for most home/desktop use.
There is a specification called "Total Bytes Written" (TBW) but I do not see one listed for the 510.
However, here's an example of a near worst case scenario for most people:
You have a 120GB SSD with 128GB of NAND.
The NAND has only 3000 cycles.
The write amplification is 1.2x.
That means you would have to write about 20,000GB worth of date before you use up the reserve space.
The majority of users write less than 10GB a day to their disk.
That means your SSD is good for at least 5 years and again is worst case for most people.
Using more conservative numbers, you are looking at around 7-10 years before you run out of reserve space. However, the worn-out pages are still readable AND the rest of the drive is still usable. Basically, it isn't that big of an issue for most home users.
SSD use NAND memory which are non-volatile.
The program that you mentioned is probably the Intel SSD Optimizer, which is part of the Intel SSD Toolbox. It is recommended for use once a week. The optimizer is most useful with OSs that do not natively support the TRIM command, such as XP and Vista, but is good insurance when used with OSs that support TRIM, such as Windows 7. The SSD Optimizer performs the same function as the TRIM command, but done manually by the user.
The TRIM command, or really it's function, is not a fix for SSDs, but simply a function that SSDs need to perform at their optimum level over time. OSs have been written with the needs of HDDs in mind, but SSDs function differently and OSs are only recently taking their needs into account. For example, SSDs do not have sectors, cylinders, or tracks as HDDs do, so that model of a storage device does not apply to SSDs. SSDs do not need de-fragmentation as HDDs do, and de-fragmenting an SSD actually has a negative affect on it. I'm sure that in the future we will see the file systems of OSs change to fit the way SSDs function, which will likely add to their performance.
Thank You. In terms of archive of data, it seems like just when I've just found the perfect solution by digitizing the media of human interaction of sight and sound, so that there is no degredation to the 'nth' playback, I find out that the media I store it on won't last past 5 years. Even the dye in the CDs and DVDs and Blu-rays, being orgainic have a shelf-life, like every other organic compound. I have books that are 40 years old, and I can still read the pages and see the colored pictures. It is a shame that this kind of burned-in information can't be kept undegraded for at least as many years when translated down on a nano level.
I remember telling my friends years ago that someday data would be stored in a solid crystalline block with the data written along the crystal faces, and read holographically from each axis that could produce a face in that crystalline structure. Back then, hard drives spun slower, and had less memory than a 2Gig memory stick. I still have to lament so short a storage life in today's media, without drastic intervention by rewriting to new media. The memory leveling reminds me of early television when they still used videcon tubes. They engineered the cameras so that the scan dithered ever so slightly just so that if you had a bright spot in the picture it wouldn't prematurely age the phosphor it hit in one spot, which, otherwise eventually led to a 'burn' or black dot in the picture where the phosphor had been damaged. A lot of local stations, in order to save money , would continue to run these cameras with damaged tubes, hoping it wouldn't be noticed, until either the damage started becoming too great, or they accumulated the money to take the camera offline and replace and realign the tube. Back them, before widespread cable-TV, the rabbit ears and shadows from reception's multipath ghosting was more objectionable than some little dot, anyway.
As with everything in the world of computers, there are endless details, and Duckie gave you the overview of SSD longevity. The NAND cells of an SSD have a finite number of write operations they can endure, before they no longer will be writeable. Duckie's numbers are conservative, I have seen the number of writes as 10,000. While the write operations may be used up, the NAND cells may still be read many, many more times. Not permanent, but better. CDs and DVDs may last 50 to 100 years, if you can find a device to read them 100 years from now, which would include all storage media we use today.
Except for one that is, that being the printed paper page, which has yet to be beaten for longevity, given the right conditions of storage, and ease of being read.
Than you. It is just this type of frankness of the current recordable media state-of-the-art that is needed for risk management for an archive that either is universally important to keep, or a personal archive that is important - if nowhere else - in the mind of the archivist, to keep. You simply have to weigh the risks versus importance of the data to develop a system of successive regeneration of the data on same or current media (down the road). Even then, you have to wait for the 'big boys' to battle it out to see who's left standing as to the standard for the new media that emerges as the latest/greatest.