SSD data retention for archived drives

[Jensen Ackles]

Several industry players flash memory have come together in an alliance to promote the creation of a standard for SSD. I am working for big corporation. I got some threads which says that the data retention ability for a ssd drive is 10 years. I think this might not be useful for very large database kept for long time. What is done more in this part of data retention. I think this might be increased.

[Odalis]

To recall, retention is the time there can be some issue with data without loss which is much needed to taken in point. Obviously by that you might need to work with the limitations of reading and writing on the flash memory. The SSDA counting on the optimizations introduced by Windows such as the recognition of a SSD will disable the automatic defragmentation or the Trim function.

[Florence_USA]

SSDs have a different characteristic compared to traditional hard drives the cells wear out over a long time by continuous usage. This gradually reduces the data retention time and at the end you need to throws this. Each cell is specified for a maximum number of erase. According to JEDEC standards, this maximum number should be set so that the retention time is guaranteed to at least ten years for a cell.

[Cruzz]

Yes it is right that the data retention ability for a SSD is 10 year. The MLC cell in the ssd has the the maximum number of formatting toleration which you can is 10,000, against 100,000 for SLC-type memory. SSDs typically are stored for long time reaching their limits. Moreover, the number of cycles incurred decreases with the fineness of engraving. The MLC does not support such 34Nm more than 3000 cycles.

[Carlos16]

One of the problem in the same is that our operating systems is not adapted to this mode of operation. It is therefore sometimes often rewrite some logical sectors, while others are untouched. The memory controllers implement the SSD algorithms therefore called as wear-leveling which will distribute the records across the surface of SSD to ensure that some cells do not wear out prematurely.

[Regina]

One thing you must know that the mode of operation mentioned above tends to reduce the write performance. Each time, the controller must go descriptors enabling the best adoption to accommodate the data. This drop in performance is particularly noticeable on SSD based MLC and small random write operations. To avoid excessive degradation of performance it is common that the DSS is equipped with a cache.

[Bon-Z]

But the wear-leveling has another problem which is causes due to fragmentation of the volume to increase time and degrade write performance. Indeed, imagine for example writing a 1 MB file on a SSD with pages of 2 KB and 512 KB blocks, here all cells are blank. If the file is written properly on the two consecutive blocks then the block becomes best part for deleting files. This procedure further degrades SSD quality.